Luminous Landscape Forum
Equipment & Techniques => Medium Format / Film / Digital Backs – and Large Sensor Photography => Topic started by: KevinA on February 11, 2010, 07:56:54 am
-
What exactly does it do? As I understand you focus on the eyes, then as you tilt the camera down it holds focus on the eyes, how is that better than a choice of AF focus points? the distance from eyes to camera has hardly changed, or does it recalculate if you or subject moves forward or backwards when tilted? Is this just Hasselblads round about way of saying we can't figure out how to do multi focus points, am I doing it disservice.
Kevin.
-
What exactly does it do? As I understand you focus on the eyes, then as you tilt the camera down it holds focus on the eyes, how is that better than a choice of AF focus points? the distance from eyes to camera has hardly changed, or does it recalculate if you or subject moves forward or backwards when tilted? Is this just Hasselblads round about way of saying we can't figure out how to do multi focus points, am I doing it disservice.
Kevin.
I think it is much better than multiple AF points. It allows you to focus and simply recompose. By calculating the movements it will readjust the focus back to where you had it. It makes the whole screen an AF point basically.
Sure it doesn't compensate for subject movement, it also probably doesn't compensate for lateral camera movements either. Every system can be made to fail, I think this can be made to work very nicely too.
For me the H4D is interesting because of this. Kudos for thinking outside the box with this.
-
What exactly does it do? As I understand you focus on the eyes, then as you tilt the camera down it holds focus on the eyes, how is that better than a choice of AF focus points? the distance from eyes to camera has hardly changed, or does it recalculate if you or subject moves forward or backwards when tilted? Is this just Hasselblads round about way of saying we can't figure out how to do multi focus points, am I doing it disservice.
Kevin.
Hi Kevin,
The problem with multi point AF on medium format is that placing the points so wide from the centre of the less would very much reduce the accuracy. Especially if you wanted to use a short depth of field. You could argue at tighter apertures that the issue would be dissolved of course.
Its not that we don't 'want' to do Multi Point AF, it is just trying to find a way that would ensure the same level of accuracy towards the edge of the optics as compared to the centre.
The off-the-shelf available AF sensors do not have wide enough spaced points for MF. If you place a D3 AF sensor in an H4D50, you actually would get very little benefit even from the widest spaced points. Also the AF sensor being an optical component itself is quite physically large, so tripling up the existing sensor is not possible and we would still have the accuracy issue. Then of course you would also need to include an active LCD in the viewfinder describing which point you were using.
So right now the alternative is True Focus. I have attached two PDF's which describe how the function is used (if interesting) and one which goes a bit more in-depth on where it matters.
I hope that helps.
Best
David
-
Hi Kevin,
The problem with multi point AF on medium format is that placing the points so wide from the centre of the less would very much reduce the accuracy. Especially if you wanted to use a short depth of field. You could argue at tighter apertures that the issue would be dissolved of course.
Its not that we don't 'want' to do Multi Point AF, it is just trying to find a way that would ensure the same level of accuracy towards the edge of the optics as compared to the centre.
The off-the-shelf available AF sensors do not have wide enough spaced points for MF. If you place a D3 AF sensor in an H4D50, you actually would get very little benefit even from the widest spaced points. Also the AF sensor being an optical component itself is quite physically large, so tripling up the existing sensor is not possible and we would still have the accuracy issue. Then of course you would also need to include an active LCD in the viewfinder describing which point you were using.
So right now the alternative is True Focus. I have attached two PDF's which describe how the function is used (if interesting) and one which goes a bit more in-depth on where it matters.
I hope that helps.
Best
David
So unless the model and photographer are nailed to the ground though you are just as likely to get it unsharp on the eyes are you not.
You focus on the eyes, tilt the camera down then the model shuffles left or right an inch or two, or you lean back and it's all a waste of time right?
I thought maybe the focus on the eyes was a reference, then when you tilted down it worked out the difference then applied that with the tilt information to any new distance information. The only way I can see it being of much value as explained would be if the model was held in clamps and camera on a tripod, you would then have the advantage of being able to shoot a wet collodion for back-up.
Does it work better in practice than it looks on paper?
Kevin.
-
So unless the model and photographer are nailed to the ground though you are just as likely to get it unsharp on the eyes are you not.
You focus on the eyes, tilt the camera down then the model shuffles left or right an inch or two, or you lean back and it's all a waste of time right?
I thought maybe the focus on the eyes was a reference, then when you tilted down it worked out the difference then applied that with the tilt information to any new distance information. The only way I can see it being of much value as explained would be if the model was held in clamps and camera on a tripod, you would then have the advantage of being able to shoot a wet collodion for back-up.
Does it work better in practice than it looks on paper?
Kevin.
Kevin,
Stinging sarcasm aside...
I thought maybe the focus on the eyes was a reference, then when you tilted down it worked out the difference then applied that with the tilt information to any new distance
AF systems don't work like that, so unless you refocused again after tilting there would be no way of knowing the 'new' distance. This would be a very slow way of working.
But you are correct in thinking that that the change in the optimum focus point is calculated based on tilt information only. Which leads me to my second point...
There is always an optimum point of sharpness in an image. Without Truefocus your focus error would be greater. With Truefocus the error is reduced.
Lets say you are focussing on the eyes of a model, portrait orientation.. when you recompose, the optimum focus point will shift further back a little. With True focus it is maintained.
Both scenarios still can't account for distance changes, the margin for error using True Focus though, is reduced.
David
-
The problem with multi point AF on medium format is that placing the points so wide from the centre of the less would very much reduce the accuracy. Especially if you wanted to use a short depth of field. You could argue at tighter apertures that the issue would be dissolved of course.
David,
Hasselblad must be applauded for even trying to address this autofocus issue. I can't see that it would work for me, but maybe it would for others' style.
I was shooting a job last week on the road, and this focus thing again just slapped me in the face. I am amazed that more photographers don't talk about it. Maybe it's just too embarrassing that, in the year 2010, we're still dealing with basic things like keeping the subject sharp, (even when shooting on a tripod, with a non moving subject). My setup, at that time, was a location environmental portrait. I was shooting with an unnamed 35mm DSLR (5d2), tethered, on a tripod. To give you an idea about how far away, I was shooting with a 50mm f1.2, at f8, and the subject was full length, in vertical format, but there was a good deal of room around their body. So in short, the subject was pretty damn far away. You'd think, with f8, on a tripod, with a non moving subject, the autofocus sensor could nail it. I used one of the outer squares for my focus sensor, due to how my shot was locked down, and I was trying to find some constrast in her clothing for the AF sensor to grab onto. Here is the key: You CANNOT trust what you're seeing in the Viewfinder. So many times, it looks TACK SHARP, but you shoot tethered, and you lean over and peek at the monitor, and for, the focus is slightly off. To me, this issue is The Great Unspoken issue about all digital cameras -- the Viewfinder, (even with StopDownPreview pressed), does not match, at all, what's actually rendered into the file.
But the truly scary thing is: I have spent DAYS with my assistant, and we test these cameras's focus issues, sitting there, shooting tethered, looking at every frame. The scary thing is: There is no rhyme or reason sometimes. Sometimes, you shoot a frame, it's tack sharp. You shoot the next frame, everything the same, (on tripod), and it's slightly soft. It's like there's some kind of "autofocus hunting" going on, even though you can't see or hear any hunting going on. Frame after frame. I have calibrated each lens; I have thrown salt over my shoulder; I have even tried religion. But this autofocus sensor thing haunts me. (I gave up on shooting Manual Focus last year).
Back to Hasselblad, here is the problem to me: maybe I'm missing something. I see myself shooting in this way, and it seems like this:
a. Compose the frame. Lock down tripod.
b. Unlock tripod.
c. Find what you want sharp on the subject, (and hope it's contrasty).
d. Put focus sensor then and press some button.
e. Hold down on that button.
f. Unlock tripod.
g. Recompose frame.
h. SHOOT.
i. Unlock tripod.
j. Find what you want sharp on the subject, (and hope it's contrasty).
k. Put focus sensor then and press some button.
l. Hold down on that button.
m. Unlock tripod.
n. Recompose frame.
o. SHOOT.
Over and over and over.
The reason I post this question: Are other people fighting focus too, no matter what digital camera you're using? I'm no fan of LiveView, but on the 5d2, it's pretty damn amazing: double click on the zoom thingie, and you're at 10x, and you focus, and YOU KNOW YOU GOT IT, because you're see what the sensor is seeing. But I am not fast enough shooting people's fleeting facial expressions to actually use LiveView.
It just seems that, overall, the Focus Tolerance for digital is MASSIVELY smaller than for Old Timey Film. So much so that, I just wonder why more people don't fight with Focus, even using AutoFocus. I could put up a strong argument that, with the tiny tolerances of these digital files, that it's clearly time to abandon the Ground Glass, and mirrors, and simply use an Electronic Viewfinder, in order to know that you nailed it.
-
David,
Hasselblad must be applauded for even trying to address this autofocus issue. I can't see that it would work for me, but maybe it would for others' style.
I was shooting a job last week, and this focus thing again just slapped me in the face. I am amazed that more photographers don't talk about it. Maybe it's just too embarrassing that, in the year 2010, we're still dealing with basic things like keeping the subject sharp, (even when shooting on a tripod, with a non moving subject). My setup, at that time, was a location environmental portrait. I was shooting with an unnamed 35mm DSLR (5d2), tethered, on a tripod. To give you an idea about how far away, I was shooting with a 50mm f1.2, at f8, and the subject was full length, in vertical format, but there was a good deal of room around their body. So in short, the subject was pretty damn far away. You'd think, with f8, on a tripod, with a non moving subject, the autofocus sensor could nail it. I used one of the outer squares for my focus sensor, due to how my shot was locked down, and I was trying to find some constrast in her clothing for the AF sensor to grab onto. Here is the key: You CANNOT trust what you're seeing in the Viewfinder. So many times, it looks TACK SHARP, but you shoot tethered, and you lean over and peek at the monitor, and for, the focus is slightly off. To me, this issue is The Great Unspoken issue about all digital cameras -- the Viewfinder, (even with StopDownPreview pressed), does not match, at all, what's actually rendered into the file.
But the truly scary thing is: I have spent DAYS with my assistant, and we test these cameras's focus issues, sitting there, shooting tethered, looking at every frame. The scary thing is: There is no rhyme or reason sometimes. Sometimes, you shoot a frame, it's tack sharp. You shoot the next frame, everything the same, (on tripod), and it's slightly soft. It's like there's some kind of "autofocus hunting" going on, even though you can't see or hear any hunting going on. Frame after frame. I have calibrated each lens; I have thrown salt over my shoulder; I have even tried religion. But this autofocus sensor thing haunts me. (I gave up on shooting Manual Focus last year).
Back to Hasselblad, here is the problem to me: maybe I'm missing something. I see myself shooting in this way, and it seems like this:
a. Compose the frame. Lock down tripod.
b. Unlock tripod.
c. Find what you want sharp on the subject, (and hope it's contrasty).
d. Put focus sensor then and press some button.
e. Hold down on that button.
f. Unlock tripod.
g. Recompose frame.
h. SHOOT.
i. Unlock tripod.
j. Find what you want sharp on the subject, (and hope it's contrasty).
k. Put focus sensor then and press some button.
l. Hold down on that button.
m. Unlock tripod.
n. Recompose frame.
o. SHOOT.
Over and over and over.
The reason I post this question: Are other people fighting focus too, no matter what digital camera you're using? I'm no fan of LiveView, but on the 5d2, it's pretty damn amazing: double click on the zoom thingie, and you're at 10x, and you focus, and YOU KNOW YOU GOT IT, because you're see what the sensor is seeing. But I am not fast enough shooting people's fleeting facial expressions to actually use LiveView.
It just seems that, overall, the Focus Tolerance for digital is MASSIVELY smaller than for Old Timey Film. So much so that, I just wonder why more people don't fight with Focus, even using AutoFocus.
First of all what you are seeing in the viewfinder is not the widest aperture, it will be a stop or two down from that - so if you are not confident that your mirror is correctly calibrated to your viewfinder glass, even if it does look tack sharp... it may well not be.. as you have discovered. We use a laser system to calibrate the mirror to viewfinder glass. There is really very little margin of error. We are talking microns.. not 0.5mm or something. As a 5D is mass produced I don't know how long they spend on such calibrations?
You have also rather nicely pointed out the issue with widely spaced sensors.
The reason why it 'hunts' is because the AF electronics are are comparing two different images from converging optics. I don't know what the 5DMkII AF tolerance is but it will never wait for a 'perfect' 0 tolerance value from the AF electronics... It will reject 'out-of-focus' values until it gets within it acceptable ball park range. As you move to the further out AF points it will accept a much wider tolerance of error. No matter what you do, the wide AF points will never ever be as good as the centre.
There is also a difference in accuracy from Landscape to Portrait. You will probably find landscape more accurate. Try if you like.
As to point 'e' of your rather long list you do not need to hold the button down. Press once, wait for confirmation (like any camera AF system) and release.
So multi AF has it faults. True Focus is something you use consciously. There is no perfect solution yet... unless you have live view.. but that isn't much good for fast paced shooting if you want narrow depth of field.
Ideally you want a CMOS sensor comparing local contrast at any site on the sensor... but it is still slower than an optical system.
David
-
There is also a difference in accuracy from Landscape to Portrait. You will probably find landscape more accurate. Try if you like.
Just my luck. I shoot mostly verticals for advertising stuff. Thus my questions about a Hasselblad vertical grip.
This focus thing almost makes you want to mount one of those Zacuto things on the back of the LCD and abandon the viewfinder altogether.
Thanks for your thorough answer.
-
Kevin,
Stinging sarcasm aside...
I thought maybe the focus on the eyes was a reference, then when you tilted down it worked out the difference then applied that with the tilt information to any new distance
AF systems don't work like that, so unless you refocused again after tilting there would be no way of knowing the 'new' distance. This would be a very slow way of working.
But you are correct in thinking that that the change in the optimum focus point is calculated based on tilt information only. Which leads me to my second point...
There is always an optimum point of sharpness in an image. Without Truefocus your focus error would be greater. With Truefocus the error is reduced.
Lets say you are focussing on the eyes of a model, portrait orientation.. when you recompose, the optimum focus point will shift further back a little. With True focus it is maintained.
Both scenarios still can't account for distance changes, the margin for error using True Focus though, is reduced.
David
It sounds like Hasselblad have got some way there but easily could of moved on another notch. Presumedly it's a bit of trigonometry? I am no way a mathematician (I don't even have an "O" level) but it would be a short step to allow for photographer or subject forward and backward movement. If two points are recorded top and bottom from the same position every focus point on that plane would be known and the angle the camera would need to be at to achieve that. So if the camera is at an angle and focus information shows less or greater distance than referenced previously, it could easily calculate a shift of the focus plane and the new difference of focus needed for the selected "Sharp" point. It still would not allow for a subject leaning out of plane, but much more useful that everything having to be static.
I can't believe hasselblad are not working on something like this? If it is a new idea you take up, a couple of H4D's please and a selection of lenses ta very much.
Kevin.
-
Im sorry Kevin. It just isn't possible in the way you describe without making it a very laborious process.
Attached is the trig info.
The only foolproof way would be to do it Wii style, but that would mean putting an infra red sender on top of the camera and a receiver on the models head.
-
PS. I will say it is worth some kind of human anthropology experiment.
If when every one recomposes they move forward/backward by a function of the distance then it would be easy to enter an offset into the calculation.
However, it is likely that we all behave a bit differently.
-
If when every one recomposes they move forward/backward by a function of the distance then it would be easy to enter an offset into the calculation.
My idea is for Hasselblad to invent some kind of Sensor Receptor. It would be about the size of a quarter, and it could pin, or velcro, or stick to the person's clothing. Wherever that Receptor was, the H lens would focus on that. So let the subject move; let the subject run; let the subject dance -- wherever she goes, the Receptor goes with her, and the Hasselblad follows focus. Maybe it emits some kind of sound that's outside the range of human hearing; maybe it's painted some color that's not in the Crayola box -- something, anything, to make it unique, so that the Hasselblad is not fooled.
I'm kidding, but not really.
-
Im sorry Kevin. It just isn't possible in the way you describe without making it a very laborious process.
Attached is the trig info.
The only foolproof way would be to do it Wii style, but that would mean putting an infra red sender on top of the camera and a receiver on the models head.
2 readings one for top one for bottom then a third for focus point from one camera position, quicker than getting a light reading or adjusting ratios I would of thought.
Kevin.
-
Could I just ask, if someone could explain in a few words, just exactly why is digital so much harder to focus than film? I know that it is, because I have found out the hard way with my CFV back, but none of the explanations that I have seen so far seem wholly convincing.
John
-
Could I just ask, if someone could explain in a few words, just exactly why is digital so much harder to focus than film? I know that it is, because I have found out the hard way with my CFV back, but none of the explanations that I have seen so far seem wholly convincing.
John
1) The way we view files is different. Before - light box (scratched), Loupe (scratched), Film... hopefully not scratched. Now 100% view on a 30" display.
2) Pixel pitch of modern sensors is extremely small and a regular pattern. Film grains varied in size and were an irregular pattern.
3) The only point that the image is in focus on a sensor is EXACTLY on the surface. Front or behind you are moving away from the optimum focus point. Film had some inherent thickness which gave a greater latitude. For example a sensor misplaced by as little as 50 microns (0.05mm) is enough to show a focus error on an image with shallow depth of field. (If you were looking closely... see point 1!)
-
I played with the H4 yesterday and have to say I was very impressed with the true focus. I used true focus with a 120 macro, focused on the eyes and then tilted the camera down until the eyes where close to the edge and they stayed tack sharp. I also liked the fact you can shoot tethered still life, and choose a spot in the image and the camera will focus on that spot. I really think Hasselblad stepped up the game against Phase. I also think Hasselblad hype was a little over the top since this was announced last year they would have this technology.
I need to shoot a real world test, and see how the images looks against my Leaf back but this is a camera I will probably buy if the images quality is what I need.
-
Could I just ask, if someone could explain in a few words, just exactly why is digital so much harder to focus than film? I know that it is, because I have found out the hard way with my CFV back, but none of the explanations that I have seen so far seem wholly convincing.
John
Viewfinders on digital specific cameras tend to be smaller and less bright (and often without focus aids like micro prisms) ... (on average - there are always exceptions).
Some argue that slight misfocus is more obvious with digital, I'm unconvinced.
And people tend to forget the things we learned when manually focusing images. Relying on AF points and forgetting to check for themselves.
[/curmudgeon]
-
1) The way we view files is different. Before - light box (scratched), Loupe (scratched), Film... hopefully not scratched. Now 100% view on a 30" display.
2) Pixel pitch of modern sensors is extremely small and a regular pattern. Film grains varied in size and were an irregular pattern.
3) The only point that the image is in focus on a sensor is EXACTLY on the surface. Front or behind you are moving away from the optimum focus point. Film had some inherent thickness which gave a greater latitude. For example a sensor misplaced by as little as 50 microns (0.05mm) is enough to show a focus error on an image with shallow depth of field. (If you were looking closely... see point 1!)
Well said David , saved me the typing. Bottom line we NEVER blew a piece of film up 100 percent to view the best we did was maybe a 8x lupe and point 3 is right on the money.
Now having said all that I never trusted a AF system on any cam . Bottom line it can't think like we do. Center AF point is great if the composition stays in the center which doing people it never does we recompose which throws it off. MF it is even tougher. Shooting the S2 P40+ test in the studio tethering the P40+ we did pretty good even at F11 but we still missed. Now the S2 non tethered i was lucky to nail a couple. Not that it was the S2 fault per say but with a long lens like the 180 the DOF is so freaking thin to begin with and than recomposing than you just get lost. Getting critical focus on people is tough and trying to get a eyelash for a test is nuts. Lot's of hit and miss and this is at F11. I try to tether as much as possible in this types of situations. Honestly end of day bottom line i use AF to get me there than hit the clutch for manual and fine tune almost every time. Nature of the beast. The HD4 true focus looks interesting and I would like to try it sometime
-
1) The way we view files is different. Before - light box (scratched), Loupe (scratched), Film... hopefully not scratched. Now 100% view on a 30" display.
2) Pixel pitch of modern sensors is extremely small and a regular pattern. Film grains varied in size and were an irregular pattern.
3) The only point that the image is in focus on a sensor is EXACTLY on the surface. Front or behind you are moving away from the optimum focus point. Film had some inherent thickness which gave a greater latitude. For example a sensor misplaced by as little as 50 microns (0.05mm) is enough to show a focus error on an image with shallow depth of field. (If you were looking closely... see point 1!)
I don't think it's any different today than it was with film, other than with digital's instant review we attempt to shoot at wider apertures than we did with film, especially in commerce.
Commerce and film was always that fear factor of shooting a polaroid at F4, thinking yea that looks good, then the fear creeps in with the thought of well, maybe I need more of the shoulder in focus, or that model is really moving fast so I had better go up to 5.6, well let's try 8, ah to hell with it shoot it at F16 that'll do it.
With digital we preview it on the camera and/or the computer and can really get down to the bare minimum, of course when you run on the edge your always going to miss some and even if you get 200 in focus and one soft image, I can promise you the client will select that one soft image.
I think because that's the image that looks spontaneous and clients love spontaneity.
I haven't tried the hasselblad with phocus shift or whatever it's called, though I wonder how it works on moving objects.
In regards to autofocus, if you really want pinpoint sharper than sharp focus pick up a Nikon. If you can get the focus points on the subject it will be dead ass sharp, wide open, 95% of the time in almost total darkness.
Nothing I've ever used focuses like a Nikon.
A year ago we shot a round the world gig, of about 10,000 images, almost split 50/50 between a 1ds3 and a Nikon D3 and D700. I used the Nikon for low light and for anything that moved quickly, the Canon with more light and any scene that was more static and the ratio of sharp focus was about 65% Canon, 85% Nikon.
Regardless, I think focus is usually somewhat overrated. Obviously a paying client wants an image in focus and that's the goal, but do a quick tour of the galleries in New York and look at some of the world's most historically compelling images and you'd be surprised that the majority of them are soft.
Maybe it's because the subject did something unexpected, or maybe it's just because life works that way, but nobody notices a slightly soft image if it's really compelling.
For all your sharpness junkies, 10 mpx of a sharp image has a lot more detail than 60 mpx of a soft image.
Since digital, especially 20 gazillion megapixel digital all everyone seems to talk about is ultra focused, uber over sharp images, but usually the examples I see are pretty damn boring. Sharpness and extreme detail won't save a boring photograph.
In fact I believe the first 10 years of digital photography will be called the oversharp decade.
BC
-
gwhitf
You are making the process sound a lot more complicated than it really is. When you shot H did you have the AF on the user button and focus recompose? True focus works the same way but with the addition of a rapidly calculated adjustment based on the amount of tilt. What's it's doing is correcting for the inevitable back focus caused by the change in distance.
Kevin
As you (rather sniffily) pointed out the system cannot compensate for a moving model but it is still compensating.
I have been using the system for a while and it does work albeit with the caveats that others have mentioned (moving model, moving photographer).
As for film I don't think it was ever sharp, and as David says we never looked at it as closely.
My 2c
Nick-T
Edit
Forgot to mention that I did blather about true focus in the Hasselblad newsletter might be worth a read if you are trying to get your head around how it works.
-
gwhitf
You are making the process sound a lot more complicated than it really is. When you shot H did you have the AF on the user button and focus recompose? True focus works the same way but with the addition of a rapidly calculated adjustment based on the amount of tilt. What's it's doing is correcting for the inevitable back focus caused by the change in distance.
Kevin
As you (rather sniffily) pointed out the system cannot compensate for a moving model but it is still compensating.
I have been using the system for a while and it does work albeit with the caveats that others have mentioned (moving model, moving photographer).
As for film I don't think it was ever sharp, and as David says we never looked at it as closely.
My 2c
Nick-T
Edit
Forgot to mention that I did blather about true focus in the Hasselblad newsletter might be worth a read if you are trying to get your head around how it works.
So Nick as I mentioned the scenario above focus on the eye's than recompose by adjusting down ( given a vertical shot)and it calculates that tilt down. Hmmm
-
Hi all,
It seems to me that True Focus is a step forward, and bravo to Hasselblad to do that. And it also seems to me that this is achieved by software...calculating tilt, and applying it to each specific lens which the camera already knows about is just maths. So my question is would True Focus be available as a firmware upgrade to the earlier H cameras?
-
Hi all,
It seems to me that True Focus is a step forward, and bravo to Hasselblad to do that. And it also seems to me that this is achieved by software...calculating tilt, and applying it to each specific lens which the camera already knows about is just maths. So my question is would True Focus be available as a firmware upgrade to the earlier H cameras?
You'd need a tilt/yaw sensor..
-
I attended the H4d-40 launch in los angeles yesterday...and I came away with perhaps a bit of a yawn over the new true focus system. It worked about half the time seemingly, and I'm not sure whether this was due to the use of the 28mm (the rep seemed to think this lens would show the greatest effect...which I was skeptical about...but). Yes it did work, but I think I came away with the conclusion that the "reposition error" is not really as great as has been discussed, and at even f/5.6 we could barely tell the difference between the TF version and using the standard autofocus. (after several attempts I finally suggested trying it at f4, and then the difference, while still subtle, was at least noticeable. Again, perhaps the issue was the lens used.
On the other hand...i was quite surprised by the low noise this new sensor is capable of even at 1600. While the rep was talking, I managed to squeeze off a few handheld shots at high ISOs and I found the results a decent step up (first thumbnail) from my 31. I compared these shots with some 1ds Mark3 (2nd thumbnail) frames shot under similar lighting conditions (mix of incandescent and warm flourescents)... and found the 40 to be just as good at 100%. I realize there are better FF35mm cams out there now in low light, but the current 1ds is no slouch, and for a MF camera to hit this level I think is good news. At 50%, in-focus areas show not a trace of noise (third thumbnail), and I think the color rendering is really pretty good given the mix of lighting. I'd feel very comfortable with this ISO setting if I needed it in a pinch.
Barry
[attachment=20215:h4d_40.jpg]
[attachment=20213:1dsmark3.jpg]
[attachment=20216:50percent_iso1600.jpg]
-
I attended the H4d-40 launch in los angeles yesterday...and I came away with perhaps a bit of a yawn over the new true focus system. It worked about half the time seemingly, and I'm not sure whether this was due to the use of the 28mm (the rep seemed to think this lens would show the greatest effect...which I was skeptical about...but). Yes it did work, but I think I came away with the conclusion that the "reposition error" is not really as great as has been discussed, and at even f/5.6 we could barely tell the difference between the TF version and using the standard autofocus. (after several attempts I finally suggested trying it at f4, and then the difference, while still subtle, was at least noticeable. Again, perhaps the issue was the lens used.
On the other hand...i was quite surprised by the low noise this new sensor is capable of even at 1600. While the rep was talking, I managed to squeeze off a few handheld shots at high ISOs and I found the results a decent step up (first thumbnail) from my 31. I compared these shots with some 1ds Mark3 (2nd thumbnail) frames shot under similar lighting conditions (mix of incandescent and warm flourescents)... and found the 40 to be just as good at 100%. I realize there are better FF35mm cams out there now in low light, but the current 1ds is no slouch, and for a MF camera to hit this level I think is good news. At 50%, in-focus areas show not a trace of noise (third thumbnail), and I think the color rendering is really pretty good given the mix of lighting. I'd feel very comfortable with this ISO setting if I needed it in a pinch.
Barry
[attachment=20215:h4d_40.jpg]
[attachment=20213:1dsmark3.jpg]
[attachment=20216:50percent_iso1600.jpg]
Hi Barry,
What you surmise is correct, that it is an refinement of an already very accurate AF system. As I said earlier at greater DOF, you could argue the correction is not needed, but there is always an optimum point of focus and TF helps to achieve this at shallow DOF.
I am really glad you liked the results from the sensor. It is indeed a step forward.
David
-
You'd need a tilt/yaw sensor..
Good point...though would be nice to have some backward compatibility with a small clipon device for older cameras.
-
I would expect the system to be demonstrated with a very long lens ?
Portrait photographers have to focus on the eyes and than recompose the image, and I hardly ever shoot on 28mm.
Most of my work is done between 110-200mm.
-
Hi Frank,
If you do the trigonometry then with longer lenses and distances, the True Focus correction is very very small. Almost nothing!
David
-
Are other people fighting focus too, no matter what digital camera you're using?
Yes, especially with the 5DII (and 5D and D2X and D200). For exact focus I use live view and after nailed the focus with contrast AF, I just go along with shooting normally with the AF off, since I don't need it anymore for the same shot. (This is in the situation you described.) The 5DII is known for having a great central AF sensor and bad outer sensors, so why aren't you shooting a 1Ds3 which is supposed to have better outer AF sensors, or even better why not use a Nikon D3X which is supposed to be AF king? The one and only digital camera that always and always focussed dead on, was the Kodak 14n. Too bad it was so slow and had other issues too. But it nailed focus on any lens. No calibration nonsense at all, really amazing.
This focus issue with modern digital cameras has taken away a lot of the fun for me shooting at all. The non aliasing filter 14n again was good fun shooting, but a lot of work getting the files worked on. So, I wonder, why is it so difficult for manufaturers to make a well focussing camera? The 40D is supposed to be a far better focussing machine than the more expensive 5DII Why is that? Well anyway, we'll have to shoot what's available here and now, or just go and do something else for fun I guess.
David,
Hasselblad must be applauded for even trying to address this autofocus issue. I can't see that it would work for me, but maybe it would for others' style.
I was shooting a job last week on the road, and this focus thing again just slapped me in the face. I am amazed that more photographers don't talk about it. Maybe it's just too embarrassing that, in the year 2010, we're still dealing with basic things like keeping the subject sharp, (even when shooting on a tripod, with a non moving subject). My setup, at that time, was a location environmental portrait. I was shooting with an unnamed 35mm DSLR (5d2), tethered, on a tripod. To give you an idea about how far away, I was shooting with a 50mm f1.2, at f8, and the subject was full length, in vertical format, but there was a good deal of room around their body. So in short, the subject was pretty damn far away. You'd think, with f8, on a tripod, with a non moving subject, the autofocus sensor could nail it. I used one of the outer squares for my focus sensor, due to how my shot was locked down, and I was trying to find some constrast in her clothing for the AF sensor to grab onto. Here is the key: You CANNOT trust what you're seeing in the Viewfinder. So many times, it looks TACK SHARP, but you shoot tethered, and you lean over and peek at the monitor, and for, the focus is slightly off. To me, this issue is The Great Unspoken issue about all digital cameras -- the Viewfinder, (even with StopDownPreview pressed), does not match, at all, what's actually rendered into the file.
But the truly scary thing is: I have spent DAYS with my assistant, and we test these cameras's focus issues, sitting there, shooting tethered, looking at every frame. The scary thing is: There is no rhyme or reason sometimes. Sometimes, you shoot a frame, it's tack sharp. You shoot the next frame, everything the same, (on tripod), and it's slightly soft. It's like there's some kind of "autofocus hunting" going on, even though you can't see or hear any hunting going on. Frame after frame. I have calibrated each lens; I have thrown salt over my shoulder; I have even tried religion. But this autofocus sensor thing haunts me. (I gave up on shooting Manual Focus last year).
Back to Hasselblad, here is the problem to me: maybe I'm missing something. I see myself shooting in this way, and it seems like this:
a. Compose the frame. Lock down tripod.
b. Unlock tripod.
c. Find what you want sharp on the subject, (and hope it's contrasty).
d. Put focus sensor then and press some button.
e. Hold down on that button.
f. Unlock tripod.
g. Recompose frame.
h. SHOOT.
i. Unlock tripod.
j. Find what you want sharp on the subject, (and hope it's contrasty).
k. Put focus sensor then and press some button.
l. Hold down on that button.
m. Unlock tripod.
n. Recompose frame.
o. SHOOT.
Over and over and over.
The reason I post this question: Are other people fighting focus too, no matter what digital camera you're using? I'm no fan of LiveView, but on the 5d2, it's pretty damn amazing: double click on the zoom thingie, and you're at 10x, and you focus, and YOU KNOW YOU GOT IT, because you're see what the sensor is seeing. But I am not fast enough shooting people's fleeting facial expressions to actually use LiveView.
It just seems that, overall, the Focus Tolerance for digital is MASSIVELY smaller than for Old Timey Film. So much so that, I just wonder why more people don't fight with Focus, even using AutoFocus. I could put up a strong argument that, with the tiny tolerances of these digital files, that it's clearly time to abandon the Ground Glass, and mirrors, and simply use an Electronic Viewfinder, in order to know that you nailed it.
-
I would expect the system to be demonstrated with a very long lens ?
Portrait photographers have to focus on the eyes and than recompose the image, and I hardly ever shoot on 28mm.
Most of my work is done between 110-200mm.
If you shoot with portrait lens at f5.6 or f8 then there is little use of the true focus I suppose. However, when the lens is wide open the slightest error is noticeable and the TF might come in handy.
I had a chance to shoot with it this week and found the technology very naturally integrated in the camera. You don't have to think about it or press any additional button. Same as using a stand alone button for autofocusing which I do all the time anyway.
I didn't try it in continuous focus mode. I wish I had. No idea how it would behave.
Continuous is the only mode where I would prefer multiple autofocus sensors allowing me to select the point of a frame I want the autofocus to track.
My opinion concerning tripod shooting... You never have enough autofocus sensors when your priority is composition and not the convenient location of the focus point. 3,9 etc. sensors don't help. In most cases you still release the camera, focus, recompose and lock, or use manual focusing
Regards,
Alex
-
@David,
Ok,
I find that especially with the shallow DOF getting the focus perfect is more of a problem than with wide angles.
But for the movement it could be that with wide angles it's more obvious, don't know that.
I do know that I would "guess" it would be more of a problem with long lenses and shallow DOF.
F5.6 on MF is by the way giving you a very shallow DOF on 200mm.
-
I attended the H4d-40 launch in los angeles yesterday...and I came away with perhaps a bit of a yawn over the new true focus system. It worked about half the time seemingly, and I'm not sure whether this was due to the use of the 28mm (the rep seemed to think this lens would show the greatest effect...which I was skeptical about...but). Yes it did work, but I think I came away with the conclusion that the "reposition error" is not really as great as has been discussed, and at even f/5.6 we could barely tell the difference between the TF version and using the standard autofocus. (after several attempts I finally suggested trying it at f4, and then the difference, while still subtle, was at least noticeable. Again, perhaps the issue was the lens used.
On the other hand...i was quite surprised by the low noise this new sensor is capable of even at 1600. While the rep was talking, I managed to squeeze off a few handheld shots at high ISOs and I found the results a decent step up (first thumbnail) from my 31. I compared these shots with some 1ds Mark3 (2nd thumbnail) frames shot under similar lighting conditions (mix of incandescent and warm flourescents)... and found the 40 to be just as good at 100%. I realize there are better FF35mm cams out there now in low light, but the current 1ds is no slouch, and for a MF camera to hit this level I think is good news. At 50%, in-focus areas show not a trace of noise (third thumbnail), and I think the color rendering is really pretty good given the mix of lighting. I'd feel very comfortable with this ISO setting if I needed it in a pinch.
Barry
[attachment=20215:h4d_40.jpg]
[attachment=20213:1dsmark3.jpg]
[attachment=20216:50percent_iso1600.jpg]
that first thumb is a 100% crop from a 1600iso file? that looks incredible! really does...this is at least 3 or 4 stops better then anything i have seen from any other MF back....
i thought the true focus thing was made for longer lenses, for portrait guys where the point always lands on the chest? but it shows up best on superwides? i don't get that at all...but any advance is welcome i guess...
i still get the best results when i use manual focus (with the help of focus points) ...i always find AF is ALWAYS hunting...and always goes in and out even if it locks.....and people always move.....the 5DII AF is terrible compared to the rest of the camera....the worst to me is that i always feel that i am always slightly behind in timing when using AF...somehow the moment has passed when the lens is "locked on"....
-
that first thumb is a 100% crop from a 1600iso file? that looks incredible! really does...this is at least 3 or 4 stops better then anything i have seen from any other MF back....
yes it was...and I do agree it does look incredible..not just in noise...but the detail seems preserved far and away better than any Hi ISO shots I've compared it too. Here's a shot at 1600 from my H3d-31 (tungsten light only)...it's got substantial tonal banding, and clumped grain like the canon shots..not horrible...but the 40 seems to really blow this away...
[attachment=20248:h3d_31_1600A.jpg]
Back to the true focus thing...I think the implementation is really superb...seamless in my opinion. They've replaced the user button with true focus...so for those of us who AF with the user button, there really is no change to your workflow...just better focus.
I do think that if the camera was demonstrated using a 100 f2.2 at f3.2 in a close-up headshot situation, that the TF would be more evident...with the wide angle it was just weird...you'd center the face in the shot, then tilt down into heavy-distortion-land to make the effect work...just not how you'd ever use that lens (unless you were chip simons in 1985).
Barry
-
Yes, everybody here is battling focus. Bad focus ruins images taken with the sharpest lenses.
The 5DII is beloved on these forums for its image quality, ONLY. You pay $2K you get the steak ONLY, no fries.
My own experience is that the Nikon D3x offers decent focus on people, but the focus pint offset button is a real pest to use, it deactivates quickly.
A good trick in practice is to set the camera on Continuous focus, it will then track the area you focused on originally eg- face.
However AF on the Nikon has a tendency to "catch" on background high-contrast zones.
Edmund
BTW, I foundDragon Stop Motion to be an interesting tool for controlling the 5D. http://www.dragonstopmotion.com/ (http://www.dragonstopmotion.com/)
David,
Hasselblad must be applauded for even trying to address this autofocus issue. I can't see that it would work for me, but maybe it would for others' style.
I was shooting a job last week on the road, and this focus thing again just slapped me in the face. I am amazed that more photographers don't talk about it. Maybe it's just too embarrassing that, in the year 2010, we're still dealing with basic things like keeping the subject sharp, (even when shooting on a tripod, with a non moving subject). My setup, at that time, was a location environmental portrait. I was shooting with an unnamed 35mm DSLR (5d2), tethered, on a tripod. To give you an idea about how far away, I was shooting with a 50mm f1.2, at f8, and the subject was full length, in vertical format, but there was a good deal of room around their body. So in short, the subject was pretty damn far away. You'd think, with f8, on a tripod, with a non moving subject, the autofocus sensor could nail it. I used one of the outer squares for my focus sensor, due to how my shot was locked down, and I was trying to find some constrast in her clothing for the AF sensor to grab onto. Here is the key: You CANNOT trust what you're seeing in the Viewfinder. So many times, it looks TACK SHARP, but you shoot tethered, and you lean over and peek at the monitor, and for, the focus is slightly off. To me, this issue is The Great Unspoken issue about all digital cameras -- the Viewfinder, (even with StopDownPreview pressed), does not match, at all, what's actually rendered into the file.
But the truly scary thing is: I have spent DAYS with my assistant, and we test these cameras's focus issues, sitting there, shooting tethered, looking at every frame. The scary thing is: There is no rhyme or reason sometimes. Sometimes, you shoot a frame, it's tack sharp. You shoot the next frame, everything the same, (on tripod), and it's slightly soft. It's like there's some kind of "autofocus hunting" going on, even though you can't see or hear any hunting going on. Frame after frame. I have calibrated each lens; I have thrown salt over my shoulder; I have even tried religion. But this autofocus sensor thing haunts me. (I gave up on shooting Manual Focus last year).
Back to Hasselblad, here is the problem to me: maybe I'm missing something. I see myself shooting in this way, and it seems like this:
a. Compose the frame. Lock down tripod.
b. Unlock tripod.
c. Find what you want sharp on the subject, (and hope it's contrasty).
d. Put focus sensor then and press some button.
e. Hold down on that button.
f. Unlock tripod.
g. Recompose frame.
h. SHOOT.
i. Unlock tripod.
j. Find what you want sharp on the subject, (and hope it's contrasty).
k. Put focus sensor then and press some button.
l. Hold down on that button.
m. Unlock tripod.
n. Recompose frame.
o. SHOOT.
Over and over and over.
The reason I post this question: Are other people fighting focus too, no matter what digital camera you're using? I'm no fan of LiveView, but on the 5d2, it's pretty damn amazing: double click on the zoom thingie, and you're at 10x, and you focus, and YOU KNOW YOU GOT IT, because you're see what the sensor is seeing. But I am not fast enough shooting people's fleeting facial expressions to actually use LiveView.
It just seems that, overall, the Focus Tolerance for digital is MASSIVELY smaller than for Old Timey Film. So much so that, I just wonder why more people don't fight with Focus, even using AutoFocus. I could put up a strong argument that, with the tiny tolerances of these digital files, that it's clearly time to abandon the Ground Glass, and mirrors, and simply use an Electronic Viewfinder, in order to know that you nailed it.
-
Chip Simmons! Wow, I haven't heard that name in a long time. Makes me think of gelled speedlights.
Those 1600 shots do look amazing. Way better that my old P30+ or my current Aptus.
yes it was...and I do agree it does look incredible..not just in noise...but the detail seems preserved far and away better than any Hi ISO shots I've compared it too. Here's a shot at 1600 from my H3d-31 (tungsten light only)...it's got substantial tonal banding, and clumped grain like the canon shots..not horrible...but the 40 seems to really blow this away...
[attachment=20248:h3d_31_1600A.jpg]
Back to the true focus thing...I think the implementation is really superb...seamless in my opinion. They've replaced the user button with true focus...so for those of us who AF with the user button, there really is no change to your workflow...just better focus.
I do think that if the camera was demonstrated using a 100 f2.2 at f3.2 in a close-up headshot situation, that the TF would be more evident...with the wide angle it was just weird...you'd center the face in the shot, then tilt down into heavy-distortion-land to make the effect work...just not how you'd ever use that lens (unless you were chip simons in 1985).
Barry
-
David,
Is this the same sensor as used in the Leica S2? Can you give us the reference? I would like to look it up.
Thank you.
Edmund
Hi Barry,
What you surmise is correct, that it is an refinement of an already very accurate AF system. As I said earlier at greater DOF, you could argue the correction is not needed, but there is always an optimum point of focus and TF helps to achieve this at shallow DOF.
I am really glad you liked the results from the sensor. It is indeed a step forward.
David
-
David,
Is this the same sensor as used in the Leica S2? Can you give us the reference? I would like to look it up.
Thank you.
Edmund
Hi Barry,
What you surmise is correct, that it is an refinement of an already very accurate AF system. As I said earlier at greater DOF, you could argue the correction is not needed, but there is always an optimum point of focus and TF helps to achieve this at shallow DOF.
I am really glad you liked the results from the sensor. It is indeed a step forward.
David
-
Is this the same sensor as used in the Leica S2? Can you give us the reference?
apparently it's at least the same underlying design (Kodak "Truesense"):
http://tinyurl.com/ya52l2s (http://tinyurl.com/ya52l2s) (German)
KAF-40000 - http://www.kodak.com/global/plugins/acroba...ductSummary.pdf (http://www.kodak.com/global/plugins/acrobat/en/business/ISS/productsummary/FullFrame/KAF-40000ProductSummary.pdf)
-
David,
Is this the same sensor as used in the Leica S2? Can you give us the reference? I would like to look it up.
Thank you.
Edmund
it appears that Kodak doesn't list the S2 (KAF-37500) sensor on its product pages..so we don't have all the specs..although indeed it seems to have a similar design approach (6 micron pixel pitch, micro lenses) as the KAF-40000 that's in the hasselblad. It isn't the same chip (smaller format, 3:2 aspect ratio), but certainly a close cousin.
http://www.kodak.com:80/global/en/business...l?pq-path=14425 (http://www.kodak.com:80/global/en/business/ISS/Products/Fullframe/index.jhtml?pq-path=14425)
article about the truesense sensors..
http://www.kodak.com/global/en/business/IS...ath=12991/13672 (http://www.kodak.com/global/en/business/ISS/News/pressReleases/archive/2008/pr9.jhtml?pq-path=12991/13672)
-
I got to handle the H4D 40 with the 100mm 2.2 shooting wide open. I have to agree with Barry that the photographers who normally use the user-button as their focus button will find that True Focus is amazing. The images that I took of a Hasselblad representative at a distance of 1-3 meters confirm that the camera will stay focused on the face and that the chest or belly will remain out of focus. I would also like to mention that the camera felt like it had a softer mirror-slap, but that might just be me.
I did shoot some images at ISO 1600, and although I didn’t bring another camera for comparisons, I have to say, I really like the look of the grains in the Hasselblad files.
I think that Hasselblad in 2009 got a bit closer to their goal. Soon they might have a blockbuster in their hands, but there are still some small steps to take before this happens - and I doubt they are be ready to take them.
I do know that with just about every job I work on where the files are delivered as raw to the client, the client will use Photoshop CS3 or CS4 for processing. A good percentage of agencies still use CS3 or sometimes even CS2. I know this, because I frequently get the call from the agency asking why the Phase or cr2 file is only a few hundred pixels wide. All this doesn’t matter though, because Hasselblad’s FFF files don’t open in any other application then Phocus. Hasselblad has to get CS4 onboard, without waiting for CS5. Do it now, the transition is as important as the availability of Hasselblad in any rental house across America.
The second thing I would suggest to Hasselblad is to change the concept of Ultra Focus: where a back and body are matched together in order to deliver perfectly sharp focus. Of course its an ideal starting point, but what happens if the body fails. You would have to purchase a second spare body. Okay, fine, but what if I would like to buy two backs. I would have to buy 4 bodies in order to back-up my system. Why not use the same micro-focus system as Canon has implemented in their camera bodies? Instead, Ultra Focus prevents me from using my H2 body with the H4D 40 digital back. I know it’s probably not as perfect as if the combo were matched in heaven, but in an emergency, why not?
This leads me to the fact that I have two Leaf backs and 3 Hasselblad H2 bodies. If I want to upgrade to two new Hasselblad H4D 40s, I would have to buy 4 bodies and two backs. In other words, for people who upgrade into the Hasselblad system, there is an unnecessary cost exponential in comparison to Phase or Leaf. On the other hand, if you are a first time buyer, and you know that you are buying a system that will work for you, i.e. for the next 5 years, then it's a smooth ride because you don’t have any old valuable luggage you have to ditch at the side of the road.
Either way, I liked what I saw and I’m intrigued, but my dream has to be a 60mp with the same micro-lenses as the 40mp = ISO 1600
On a side note, those of my clients that still choose to shoot with digital backs are photographers who want the ultimate in image quality, which makes me think twice about the form factor of the H4D 40, since the 40mp is awfully close to the next generation of Canon 1Ds cameras. Who would choose a 40mp Hasselblad, Phase, or Leaf digital back over a DSLR at roughly the same resolution?
.e
-
I would also like to mention that the camera felt like it had a softer mirror-slap, but that might just be me.
You got that impression too? That's how I felt...but considering there's been nothing about it in the literature, I figured it was just me.
On the supplying Raw files to the client. Now this is just something I almost never do...I really don't trust my clients to know what they are doing so I always, always, always insist on mastering out the file for them into either an 8 bit or 16 bit tiff. Part of this is really about controlling image quality...and also from the business side, it's essentially a mandatory add-on fee, and "while I'm at it, should I clean up her skin a bit?" Not as cynical as it sounds. The business of high end digital is not tilted in the favor of the photographer...best to take control of as much of the work as you can, IMHO. Hasselblad's proprietary software sort of acts as a "gatekeeper" for me. Most of my clients are quite happy that they get the image ready to go. (truth be told I'm also art director and designer on much of my client work...so I probably have less resistance than others to supplying a finished image)
Barry
-
Is this just Hasselblads round about way of saying we can't figure out how to do multi focus points, am I doing it disservice?
Kevin.
Yeah probably. I think hasselblad's argument goes like this...Developing a multi-point sensor for MF would be cost prohibitive. Canon and Nikon can spread out their R&D across millions of units...but MFD's total market is in the 10's of thousands. So what they would typically do is simply purchase a sensor unit from canon, or whoever is making it for them, but the problem is that sensor would do little more than cover the center of a mfd frame (at the demo they had some sheets showing a 35mm AF sensor pattern overlaid on a MFD frame). If you look at your 35mm..you'll see that the sensors are really clustered around the center anyway, but the problem is worse with mf. I think also that when you look at how MF is typically used in commercial applications, most or at least many photographers are either manual focusing or selectively autofocusing (focus/recompose) anyway...it's not like a sports camera where your hope is that one of the sensors locks onto a basketball player. The camera's aren't being used that way...the autofocus isn't fast enough really for that application...so typically you are using a hasselblad like you've always used a hasselblad...
That said, Hasselblad's AF is incredibly accurate, better than my eyes 100% of the time. This new feature will help in the 25% of the time when focus-recompose doesn't work well..and that is an improvement.
Barry
-
I would also like to mention that the camera felt like it had a softer mirror-slap, but that might just be me.
You got that impression too? That's how I felt...but considering there's been nothing about it in the literature, I figured it was just me.
David Grover, would you comment on this, please?
Thank you,
Billy
-
@David,
Ok,
I find that especially with the shallow DOF getting the focus perfect is more of a problem than with wide angles.
But for the movement it could be that with wide angles it's more obvious, don't know that.
I do know that I would "guess" it would be more of a problem with long lenses and shallow DOF.
F5.6 on MF is by the way giving you a very shallow DOF on 200mm.
Hi Frank,
Yes, you are certainly right in saying that with shallow DOF it is much harder for perfect focus. Normally though, with a longer lens, your focus / recompose angle will actuallybe quite small, especially if you are a few metres from the subject. It then results in a very small TF correction.
Your issues probably come from other errors in the AF system or simply a model who dares to move 4mm after you have focussed!
David
-
David Grover, would you comment on this, please?
Thank you,
Billy
As far as I am aware no changes have been made to mirror mechanics.
It is probably because you read so much about 'mirror slap' here that you expect it to feel it more than there actually is!
Remember, you can also set a few ms delay on the leaf shutter firing after the mirror is UP, to help things.
D
-
Are other people fighting focus too, no matter what digital camera you're using?
Yes, especially with the 5DII (and 5D and D2X and D200). For exact focus I use live view and after nailed the focus with contrast AF, I just go along with shooting normally with the AF off, since I don't need it anymore for the same shot. (This is in the situation you described.) The 5DII is known for having a great central AF sensor and bad outer sensors, so why aren't you shooting a 1Ds3 which is supposed to have better outer AF sensors, or even better why not use a Nikon D3X which is supposed to be AF king? The one and only digital camera that always and always focussed dead on, was the Kodak 14n. Too bad it was so slow and had other issues too. But it nailed focus on any lens. No calibration nonsense at all, really amazing.
This focus issue with modern digital cameras has taken away a lot of the fun for me shooting at all. The non aliasing filter 14n again was good fun shooting, but a lot of work getting the files worked on. So, I wonder, why is it so difficult for manufaturers to make a well focussing camera? The 40D is supposed to be a far better focussing machine than the more expensive 5DII Why is that? Well anyway, we'll have to shoot what's available here and now, or just go and do something else for fun I guess.
If I was focusing then recomposing I would set my Canon to focus only when the back button is pressed, focus on the eyes then recompose.I can't see any advantage over this method with true focus if it can not adjust for subject or camera back and forth movement.
And yes I never had focus issues with a Kodak SLR/n either, Canon has been a totally different story. I hope true focus works like a dream, I just can not see it being something you could rely on to get what you need as it is. A step in the right direction, but a big enough step to be truly useful, time will tell.
Kevin.
-
Mark, you probably already know this, but did you put one of these in your 5D2?
That should help a lot in terms of making sure that what you see in the viewfinder is what shows up on the monitor. They are $40.
It makes the DOF you see through the viewfinder look more like 2.8, instead of f4. View is very slightly darker, but well worth it if you are using fast lenses with shallow DOF.
re:
http://www.bhphotovideo.com/c/product/5904...sion_Matte.html (http://www.bhphotovideo.com/c/product/590458-REG/Canon_3357B001_Eg_S_Super_Precision_Matte.html)
-
And I agree, these 1600 ISO samples look awesome.
(nice that people are finally able to focus on what this camera has to offer, instead of obsessing over what it does not ;-))
-
Hi Kevin,
The problem with multi point AF on medium format is that placing the points so wide from the centre of the less would very much reduce the accuracy. Especially if you wanted to use a short depth of field. You could argue at tighter apertures that the issue would be dissolved of course.
Its not that we don't 'want' to do Multi Point AF, it is just trying to find a way that would ensure the same level of accuracy towards the edge of the optics as compared to the centre.
The off-the-shelf available AF sensors do not have wide enough spaced points for MF. If you place a D3 AF sensor in an H4D50, you actually would get very little benefit even from the widest spaced points. Also the AF sensor being an optical component itself is quite physically large, so tripling up the existing sensor is not possible and we would still have the accuracy issue. Then of course you would also need to include an active LCD in the viewfinder describing which point you were using.
So right now the alternative is True Focus. I have attached two PDF's which describe how the function is used (if interesting) and one which goes a bit more in-depth on where it matters.
I hope that helps.
Best
David
As the issue of multipoint AF comes up regularly in almost all MF forums, perusal of the Nikon AF technical brief could clear up a lot of quries - http://www.nikon.com/about/technology/core...e/caf/index.htm (http://www.nikon.com/about/technology/core/software/caf/index.htm)
Briefly
1. The AF module is a separate optical piece, hence will have to be as large as the senor
2. Such large modules are not available off-the-shelf, hence some one has to develop it! I guess that given the low numbers of MF cameras involved (compared to 35mm DSLR) it would add quite a lot to the cost as well as to the bulk.
3. Finally as David pointed out it would require a liveview, which I believe is not feasible (as yet) with CCD
In conclusion till there is a third party Multipoint 645 AF developed and licenced, we have to live with the current single point AF with various work arounds like True-Focus!
-
@David,
To be honest I never have focus issues, only when shooting with the RZ67ProII with low key lights at the end of the day.
I've been using center focus for as long as I can remember, but I hear a lot of photographers complaining about it, maybe I'm just oldfashioned but I believe that if people could do it 20 years ago we should be able to do it now with brighter viewfinders.
However I love progress.
So I hope that it drips down to Canon, Mamiya etc.
It's always good to have something that could help.
About models moving 4mm, that's indeed not done, we use superglue and gaffer tape for that.....
-
Regarding focus - I wish for something more along the lines of a bigger brighter viewfinder than I do fancy auto focus electronics. Larger sensors and faster glass would make this possible. I have a canon 5D2 and even with all the selectable points, I miss focus more than I'd like and rely on the live view feature - but this is the same as just using a bright viewfinder only a lot slower and I can't hold the camera to my eye. So were MF manufacturers to come out with a multi-point AF, I'd probably have a similar satisfaction with it. What would be awesome is some kind of EVF with the Canon eye tracking stuff (EOS 3?) - where the camera would sense where you are looking and zoom in and out at that spot with the push of a button. But until that happens a big bright viewfinder would be just fine.
-
i really wish now that i would have gone to that demo....mostly because i can't believe the 1600 iso quality.....
i looked at some "old" P30 files and they did not look that good at 200....i am serious.....this 1600 iso combined with 40mpix file blows the canons/nikon AWAY...incl the next unannounced models.....
which is why i have to see for myself....how the hell can hasselblad go from the performance of available backs to this? the samples of binned files from the p65+ are not even close...and smaller....
raw files from the leica S (same sensor family afaik) don't compare at all...even at 320....
so how on earth is this possible?
so if anyone can offer a raw file, i would be very happy....
can the back provide this quality only out of phocus? afaik aperture supports hass files and there is no way i would ever go back to any (leaf, hass, sinar or even C1) software other then full workflow solutions (aperture, LR)....the time saved is WAY too valuable to me....so the question is: is there a lot of software magic behind this wonder and would i have to use phocus to get those results?
-
i really wish now that i would have gone to that demo....mostly because i can't believe the 1600 iso quality.....
I'm very impressed by that sample, I think the "leap" forward must be due to two things. First off the sensor is the new Kodak "True sense" design which is less noisy and has better dyes and I guess a bunch of other techy stuff. Second I know the R&D boys in Denmark have been putting a ton of work into high ISO performance (they release a free firmware update not so long ago bumping existing backs by a stop), so I guess we are seeing the results of that as well.
Nick-T
-
Not too far off topic I hope, but I was really happy with the ISO 400 shots from my ixpress 528c in single shot and they only get better with the multishot. Since I only recently aquired this back, I don't know if this was improved with firmware or not but its great and its clear Hasselblad is doing very well in this area. I am sort of hoping that this new sensor will make its way into the CF series so that I might mount it on my Rollei.
-
Doing multishot at 400ISO you will get totally rid of the color noise. What remains are grainy and very nice transitions. Absolutely lovely, not too many people know this or have even seen it.
Back on topic. The H4D40 looks promising indeed. I am sure I get a chance to give one a test ride one of these days.
-
i looked at some "old" P30 files and they did not look that good at 200....i am serious.....
I wonder a little bit about that.
The above posted ISO1600 actually reminds me a little bit of... my P21+ at ISO800.
If you take pixel pitch and resolution into account this would still be an advantage of 2 stops in favor for the H40.
Be that as it may - I am also impressed by that high ISO shot.
As to your P30: maybe try a current version of Capture One as it renders the files less coarse.
Too, the noise reduction is very much improved.
-
I'm very impressed by that sample, I think the "leap" forward must be due to two things. First off the sensor is the new Kodak "True sense" design which is less noisy and has better dyes and I guess a bunch of other techy stuff. Second I know the R&D boys in Denmark have been putting a ton of work into high ISO performance (they release a free firmware update not so long ago bumping existing backs by a stop), so I guess we are seeing the results of that as well.
Nick-T
I'll go back and see if I can retrieve the rest of the files I shot that day. I did another series of shots, again handheld there on the 4th floor of Samy's, at all ISO's from 100-1600. My impression was that there was almost no difference between the 800 and 1600 shots (a few hot pixels showed up in the 1600, and the grain got a little more ragged) but otherwise the color and detail stayed about the same, and the overall noise level seemed unaffected. And while at 100 there is slightly finer detail, and, of course, considerably less noise, surprisingly there is little difference in terms of tonality and color between the default iso and the higher gain shots. I remember the days when I'd underexpose my ixpress 132c by a stop in good light and the skintones would posterize and go grey. I'm simply shocked at what this camera can do at 4 stops underexposed in a mix of flourescent and tungsten.
Barry