Luminous Landscape Forum
Equipment & Techniques => Digital Cameras & Shooting Techniques => Topic started by: jing q on May 14, 2009, 10:53:14 am
-
just to share, I just removed the AA filter on my 5D Mk II and I see some visible improvement on the edge definition of my pictures. Will be testing more to see the effect
Take a look at the DVD near the center of the image and the magazine pages to the right of the image. There's abit of CA but each page has well defined edges.
(http://farm4.static.flickr.com/3366/3530835668_b774d5ba68_o.jpg)
-
just to share, I just removed the AA filter on my 5D Mk II and I see some visible improvement on the edge definition of my pictures. Will be testing more to see the effect
Take a look at the DVD near the center of the image and the magazine pages to the right of the image. There's abit of CA but each page has well defined edges.
(http://farm4.static.flickr.com/3366/3530835668_b774d5ba68_o.jpg)
BTW this was opened quickly in ACR and processed with a default 25 sharpening setting...
-
how did you remove the aa filter?
can you upload some pics at ISO 100 f8 or similar?
thx
-
There's also moire in the curled magazine pages. What do you normally shoot with your 5DII that would require you to remove the AA filter?
-
There's also moire in the curled magazine pages. What do you normally shoot with your 5DII that would require you to remove the AA filter?
I shoot big groups with lots of details.
the mushiness issue really crops up when resized to about 4ft by 6 ft.
I know alot of people have questioned the point of removing the AA filter but for me I feel that it's a good investment.
I have no issues with moire, and I don't shoot alot of frames so removing moire is not a big deal for me.
-
Do you have before and after images of the same subject?
nope.
will post some when I can borrow a friend's unaltered 5dMkii
-
I shoot big groups with lots of details.
the mushiness issue really crops up when resized to about 4ft by 6 ft.
I know alot of people have questioned the point of removing the AA filter but for me I feel that it's a good investment.
I have no issues with moire, and I don't shoot alot of frames so removing moire is not a big deal for me.
Just wondering if you had tried other RAW processors before resorting to this. Also, did you have a clear glass replacement installed in it's place?
As an aside to others thinking along these lines, it should be performed by a qualified technician. The AA is not user removable and requires disassembly. A clear replacement is strongly advised. The sensor itself is delicate and the dust removal feature is built into the filter. Cleaning the sensor itself can quickly lead to an expensive disaster.
-
http://www.maxmax.com/IRCameraConversions.htm (http://www.maxmax.com/IRCameraConversions.htm)
These guys do it right. Other places are somewhere in-between worthless and barely adequate. But he did say he had it removed, which means he went someplace to get it done.
-
Just wondering if you had tried other RAW processors before resorting to this. Also, did you have a clear glass replacement installed in it's place?
As an aside to others thinking along these lines, it should be performed by a qualified technician. The AA is not user removable and requires disassembly. A clear replacement is strongly advised. The sensor itself is delicate and the dust removal feature is built into the filter. Cleaning the sensor itself can quickly lead to an expensive disaster.
done with Maxmax.so they replaced the glass of course.
no it's not a raw processor issue.loss of detail is noticable on upscaling beyond a certain dimension, which no amount of sharpening can bring back unfortunately
so I figured the tradeoff was no big deal for me.
although my prism did come back chipped from the shipment. Maxmax said it wasn't their fault. told me to call UPS.
oh well.shrugs. just a small chip. abit irritating that's all.
-
although my prism did come back chipped from the shipment. Maxmax said it wasn't their fault. told me to call UPS.
oh well.shrugs. just a small chip. abit irritating that's all.
Not so sure I'd buy their claim that it wasn't their fault. I can see a lot of other things happening with the camera in shipping, but a chipped prism? Even if it wasn't done during the AA removal procedure, and did, indeed happen in transit, then their packing comes into question.
Did the box look like it took a substantial hit?
-
done with Maxmax.so they replaced the glass of course.
no it's not a raw processor issue.loss of detail is noticable on upscaling beyond a certain dimension, which no amount of sharpening can bring back unfortunately
so I figured the tradeoff was no big deal for me.
although my prism did come back chipped from the shipment. Maxmax said it wasn't their fault. told me to call UPS.
oh well.shrugs. just a small chip. abit irritating that's all.
By the prism, are you referring to the pentaprism for the viewfinder? If so, there would be no reason for the tech to disturb it in any way while working with the sensor assembly and the top cover would have to be smashed to damage it.
The only other scenario I can think of would be if the prism was loose in it's moorings and gets chipped rattling around, which would probably cause other damage as well. It's surrounded by plastic and circuit boards. It would take significant G forces to knock it loose, but not impossible, and the thing would rattle like a castanet.
All in all, a bit odd.
-
Comparison photos at http://www.maxmax.com/hot_rod_visible.htm (http://www.maxmax.com/hot_rod_visible.htm) do not, in fact, appear to be comparable. When they are downloaded and examined carefully in Photoshop it appears there was a focus difference between the modified and standard 5D. The standard 5D was focused slightly closer and has a forground that is sharper than the modified "Hot Rod" sample. With respect to the apparent sharpness advantage they suggest in the parts of the image they've highlighted for comparison we may, in fact, be seeing nothing but a focus shift of the lens.
(In addition to the foreground being sharper in the stock 5D, the stock 5D features are slightly larger -- which is consistent with the lens being focused closer. Focusing closer has the effect of increasing the focal length. You can precisely measure the distance between features on the frames to see this slightly higher magnification in the stock 5D sample.)
Paul
www.PaulRoark.com
-
Comparison photos at http://www.maxmax.com/hot_rod_visible.htm (http://www.maxmax.com/hot_rod_visible.htm) do not, in fact, appear to be comparable. When they are downloaded and examined carefully in Photoshop it appears there was a focus difference between the modified and standard 5D. The standard 5D was focused slightly closer and has a forground that is sharper than the modified "Hot Rod" sample. With respect to the apparent sharpness advantage they suggest in the parts of the image they've highlighted for comparison we may, in fact, be seeing nothing but a focus shift of the lens.
(In addition to the foreground being sharper in the stock 5D, the stock 5D features are slightly larger -- which is consistent with the lens being focused closer. Focusing closer has the effect of increasing the focal length. You can precisely measure the distance between features on the frames to see this slightly higher magnification in the stock 5D sample.)
Paul
www.PaulRoark.com
You're assuming the shots were taken from exactly the same location, on a tripod. What's more likely is that they used the same camera, on different days, from roughly the same location. If that's the case, the focus difference might just due to this.
-
You can pull up the information from the raw files and see that they were shot with different cameras. The dates are way different, but this appears to be due to one of the cameras being inaccurate. The location of parked vehicles indicates it was taken close to the same time.
Paul
www.PaulRoark.com
-
You can also copy and paste the central section of the stock 5D image over the modified image and see that the stock image is slightly more magnified, indicating the lens was focused closer. (Put the copied layer at 50%.) A slight movement of the tripod would not affect the relative sizes of the distant objects significantly. At 200% the misalignment of the images is obvious.
These are not comparable images. The test images on the web page appear to give us NO information about how much sharpness we can gain from removal of the AA filter.
Paul
www.PaulRoark.com
-
These are not comparable images. The test images on the web page appear to give us NO information about how much sharpness we can gain from removal of the AA filter.
Great analysis! So how much sharpness / edge definition / other goodness would we gain from removing the AA filter?
Thanks.
-
How much, if any, sharpness we can gain from removing the AA filter needs better testing than I've seen so far. It may be a different issue than the sharpness difference between a camera that has an AA filter and one that is designed not to have one. I can imagine, for example, that there could be a software component in the processing -- pre-raw file -- that affects the bottom line. Then, of course, the quality, thickness, index of refraction and other characteristics of the replacement for the AA filter have to be very closely matched to the OEM design.
I have a 5d2 and mostly shoot landscapes and other subjects that would not appear to benefit from the AA filter. So, if there is a way to squeeze more sharpness out of the 5d2 I'm certainly interested.
However, so far, I simply have not seen the rigorous testing that is going to be needed to show a clear net benefit. A piece of glass over the sensor is a very important optical component of the system. A net negative effect is easy to imagine if the modification is not very well designed and executed. (Some of the Canon mount conversions that look great in casual tests but do not allow the, e.g., Zeiss Contax, lens to focus to infinity come to mind.)
Paul
www.PaulRoark.com
-
However, so far, I simply have not seen the rigorous testing that is going to be needed to show a clear net benefit. A piece of glass over the sensor is a very important optical component of the system. A net negative effect is easy to imagine if the modification is not very well designed and executed. (Some of the Canon mount conversions that look great in casual tests but do not allow the, e.g., Zeiss Contax, lens to focus to infinity come to mind.)
Agreed. I'm still on the fence to remove the AA from my D300 by way of relatively inexpensive upgrade. But I have yet to see compelling before/after examples for my way of shooting (landscapes, cities etc - where I would like to gain acuteness / detail, particularly in the infinity focal range).
-
Moire and other aliasing problems can crop up when you least expect it. At least with an MF like the Leaf there are anti moire tools built into Leaf Capture to handle the occasional times this happens.
I nearly blew a very large job because of Moire on the Aptus 75, only Leaf Capture could save these critical images, none of the other suggested solutions worked.
At 33Mp aliasing effects are far less noticable than on the Aptus 22, so I guess, size of pixels is important, maybe the 5D2 will be OK it will be interesting to hear about results further down the line.
I'd say to test thoroughly before a critical shoot, and watch out for the wonderful 'greek patterns', they are impossible to get rid of, unlike regular moire. When I get those patterns it usually means 4 hours in photoshop rebuilding part of the image with a paintbrush. I've even had this on the 1DsII with a very sharp lens like the 90 TS-E, so I guess it is possible with the 5D2 especially if the AA filter is removed.
-
I have an old Kodak SLR/n which does not have a AA filter. The moire was such a problem in a few months I bought a Canon 1DsmkII, problem solved. Yes penny for penny sharper but not enough to go through all that pain again. As for moire removal techniques I tried most non removed the luminance moire one bit.
Often you would end up blurring sections to remove the moire so much it lost more detail than a filter would have. Large groups of people will get moire in clothing and hair, not pretty when blown up big, I think it was a bad idea for your subject. Sounds like the P60 was made for your subject or LF film.
Kevin.
-
I finally had the chance to do abit of testing with a friend's 5D.
First thing's first:
it's near impossible to get everything aligned.
Slight variance in positioning due to things like the release plate not being exactly at the same spot...The two files are cropped down abit to match up in terms of positioning.
Next is colour temperature variance...not sure if the colours between different units of 5D mkII cameras are different but one camera was definitely more magenta, one was more on the green side. The modification of the glass definitely plays a role here. The difference in colour cast probably has an effect in the contrast of the image.
focusing variance was also an issue. out of the 10 different series of shots we did this was the most close series of pictures we had in terms of focus from two different cameras using the same lens
85mm f1.8 wide open
focus is supposed to be right smack in the centre of the picture, on the right edge of the grey speckled pillar.
To compensate for focus error I guess you'll have to compare detail of the closest object to the furthest object.
Nonetheless, certain interesting details can be observed when zoomed in to 400%, visible at 200% too.
Would love to do a stricter test to see the effect of this modification, when I have the time
full size (slightly cropped) jpegs here
Processed in ACR, with sharpness set at 10 radius 1.0, noise reduction at 2
http://superhyperreal.com/compare1.jpg (http://superhyperreal.com/compare1.jpg)
http://superhyperreal.com/compare2.jpg (http://superhyperreal.com/compare2.jpg)
-
It's near impossible to get everything aligned. [...] Focusing variance was also an issue.
Obviously. The two images you posted are focused at very different distances.
... using the same lens, 85 mm f/1.8 wide open.
Huh!?
You want to demonstrate the effects of a removed AA filter, and for the test shots you're using a fast lens wide open? Are you nuts?
Nonetheless, certain interesting details can be observed when zoomed in to 400 %, visible at 200 % too.
In those two images, actually nothing interesting can be observed ... except that the photographer was unable to focus at the same distance twice.
... watch out for the wonderful 'greek patterns', they are impossible to get rid of, unlike regular moire. When I get those patterns it usually means four hours in Photoshop rebuilding part of the image with a paintbrush.
After all, there's a reason why digital cameras have AA filters.
-- Olaf
-
removed.
-
I put both cameras on the same tripod. Huh?
Why are you huh-ing? When the shooting distance is the same then the focus distance should be the same, too, or the image won't be comparable. Isn't that obvious?
I want to demonstrate the effects of a removed AA filter, and I used a lens wide open? Am I nuts? I must be, after all that's just how I use my camera. Sorry it didn't fit in with your idea of how I should use my camera.
Okay, let's see ... you are using your lenses mostly wide open, you cannot focus consistently, and you are less than happy with the sharpness in your images at high magnification---so you got your AA filter removed ...
Post some information to share and you get assholes talking shit.
Actually, it's misinformation you're sharing. You should keep that for yourself rather that putting bees in the bonnet ...
-- Olaf
-
Hmm, any other pictures you can share? How do you like it after the change. To be honest I'm a bit disappointed with the results of the before/after test. I keep telling myself that the 5DII would have much higher image quality without the AA filter, but your comparison shot doesn't really show what I was hoping to see. Perhaps I'm heading to Sony land.
To frame my comments, I switched from Canon to Leica a few years ago and ascribed a lot of the difference to the absence of the AA filter on the DMR. I'm sure that contributed to the look, but perhaps not as much as I thought.
Perhaps to convey to us the difference that you are seeing, can you tell us the difference in capture sharpening that you are using. On the Canons I needed quite a lot, less so on the Nikon D2x and just a hint of LCE on the Leica.
-
Hi,
AFAIK it is recommended that the image is sharpened with a high amount (like 300) and small radius (like 0.2 or 0.4) to compensate for the AA-filter. IMHO it would be more adequate to compare the images with optimal sharpening.
The effect of the AA-filter would probably be most significant at medium apertures as very few lenses can achieve optimal sharpness fully open.
Best regards
Erik
I finally had the chance to do abit of testing with a friend's 5D.
First thing's first:
it's near impossible to get everything aligned.
Slight variance in positioning due to things like the release plate not being exactly at the same spot...The two files are cropped down abit to match up in terms of positioning.
Next is colour temperature variance...not sure if the colours between different units of 5D mkII cameras are different but one camera was definitely more magenta, one was more on the green side. The modification of the glass definitely plays a role here. The difference in colour cast probably has an effect in the contrast of the image.
focusing variance was also an issue. out of the 10 different series of shots we did this was the most close series of pictures we had in terms of focus from two different cameras using the same lens
85mm f1.8 wide open
focus is supposed to be right smack in the centre of the picture, on the right edge of the grey speckled pillar.
To compensate for focus error I guess you'll have to compare detail of the closest object to the furthest object.
Nonetheless, certain interesting details can be observed when zoomed in to 400%, visible at 200% too.
Would love to do a stricter test to see the effect of this modification, when I have the time
full size (slightly cropped) jpegs here
Processed in ACR, with sharpness set at 10 radius 1.0, noise reduction at 2
http://superhyperreal.com/compare1.jpg (http://superhyperreal.com/compare1.jpg)
http://superhyperreal.com/compare2.jpg (http://superhyperreal.com/compare2.jpg)
-
Hi,
I don't think that Sony is any better than Canon, Nikon or anything else. I have a Sony Alpha 900, and it is a good camera but I don't see any magic around it.
In order to achieve optimum sharpness we need to put the camera on a steady tripod, adjust autofocus for each lens, use mirror lockup and cable release/self timer and use optimal aperture.
Best regards
Erik
Hmm, any other pictures you can share? How do you like it after the change. To be honest I'm a bit disappointed with the results of the before/after test. I keep telling myself that the 5DII would have much higher image quality without the AA filter, but your comparison shot doesn't really show what I was hoping to see. Perhaps I'm heading to Sony land.
To frame my comments, I switched from Canon to Leica a few years ago and ascribed a lot of the difference to the absence of the AA filter on the DMR. I'm sure that contributed to the look, but perhaps not as much as I thought.
Perhaps to convey to us the difference that you are seeing, can you tell us the difference in capture sharpening that you are using. On the Canons I needed quite a lot, less so on the Nikon D2x and just a hint of LCE on the Leica.
-
Why are you huh-ing? When the shooting distance is the same then the focus distance should be the same, too, or the image won't be comparable. Isn't that obvious?
Okay, let's see ... you are using your lenses mostly wide open, you cannot focus consistently, and you are less than happy with the sharpness in your images at high magnification---so you got your AA filter removed ...
Actually, it's misinformation you're sharing. You should keep that for yourself rather that putting bees in the bonnet ...
-- Olaf
there is variation in focusing distance when you are comparing 2 different camera bodies, even though they're the same camera model.
Also, since one camera has been modified, it is hard to say whether the change in filter glass has an effect on the focusing distance.
Call me paranoid but I believe using live-view is a better way of focusing for a test like this.
Taking the same lens at the same focusing distance and putting it on another camera body will introduce error in focusing. You should give it a try to see what I mean.
I focused through live-view for each camera at 10x magnification to get as exact as possible.
Before you diss me I suggest you try it yourself to see that it's not as simple as taking a lens off and putting it on another body at the same focusing distance.
Before you judge someone you should consider looking at the person's work too. My website is in my info page.
If you don't like the "misinformation" perhaps you'll like to STFU?
-
Hi,
AFAIK it is recommended that the image is sharpened with a high amount (like 300) and small radius (like 0.2 or 0.4) to compensate for the AA-filter. IMHO it would be more adequate to compare the images with optimal sharpening.
The effect of the AA-filter would probably be most significant at medium apertures as very few lenses can achieve optimal sharpness fully open.
Best regards
Erik
hi Erik
trust me I already do that (optimal sharpening)
my concern with DSLRs in the first place is to do with pixel smearing. You can sharpen an original file to bring out some detail, but the fact remains that in the first place some detail is lost when the photograph is taken
This becomes even more obvious when the image starts getting resized larger and larger. When sharpened, alot of the edges are abit muddy.
I'm basing this comparison between capturing with a medium format digital back (two different backs actually...a hassy and a leaf) and a DSLR.
I will probably try with a medium aperture when I have more time. First thing's first, trying with a wide aperture on a very narrow fous point was to see the effect on a specific spot on the image, together with the effect the modification has on the "texture" of the picture and the pixels, and to see if there is an overall increase in resolution
-
Hmm, any other pictures you can share? How do you like it after the change. To be honest I'm a bit disappointed with the results of the before/after test. I keep telling myself that the 5DII would have much higher image quality without the AA filter, but your comparison shot doesn't really show what I was hoping to see. Perhaps I'm heading to Sony land.
To frame my comments, I switched from Canon to Leica a few years ago and ascribed a lot of the difference to the absence of the AA filter on the DMR. I'm sure that contributed to the look, but perhaps not as much as I thought.
Perhaps to convey to us the difference that you are seeing, can you tell us the difference in capture sharpening that you are using. On the Canons I needed quite a lot, less so on the Nikon D2x and just a hint of LCE on the Leica.
Hi Gandalf
I recently used the 5DmkII and a H3DII on a beauty shoot. I have to say that I was pleasantly surprised at the look of the 5DmkII.
Usually it takes a round of sharpening to bring out the texture of the skin, but this time the image felt more "3 dimensional" right away upon capture.
To put it in context, when shooting with my medium format digital backs I rarely see a need for sharpening the image. Better microcontrast for sure.
With a DSLR sharpening is always necessary.
After the modification the amount of sharpening I use has decreased.
On a pixel level, edges are more distinct
I always felt that the difference between my MF files and DSLR files was that the DSLR image was always abit flatter. I attribute this to the way that pixels on the edges tend to blend into each other abit.
I believe that's why also you see a difference with your leica
Overall I was hoping that the effect would be more prominent but I am aware that the modification removed only one AA filter, (the horizontal one?) whereas the other AA filter (vertical) is stuck onto the sensor itself. Maybe that's why the Leica has a more distinct look?
But for me every single bit of resolution counts in the kind of work I do which is detail intensive, alot of small edges and details. It's abit of a psychological effect too seeing the image straight out of the camera looking nice and crisper to a degree (slightly better microcontrast)
To clarify, I'm speaking from my experience using a 1Ds MkII and a 5D MkII
Hopefully I have more examples to share in the future!
-
Taking the same lens at the same focusing distance and putting it on another camera body will introduce error in focusing. You should give it a try to see what I mean.
Unlike you, I know what you mean. I wasn't talking about apparent focus distance on the lens' distance scale but real focus distance in the final images. And the real focus distances in your test images above are significantly different---no matter how hard you tried got get them right. Just open your eyes and look at them, and you'll see what I mean!
I focused through live-view for each camera at 10x magnification to get as exact as possible.
If that really is the best you can get then removing the AA filter was a complete waste of time.
Before you judge someone you should consider looking at the person's work too.
Your work---and the quality thereof---has nothing to do with your two test images above. Those are just crap.
Hopefully I have more examples to share in the future!
Hopefully, indeed. I am looking forward to them. And put them up for us to see with no capture sharpening applied, please!
-- Olaf
-
you win, you win. Your logic is infallible.
I'm going to stop here with you.
-
Thanks Jing, that was very helpful. The Leica is basically an Imacon/Hasselblad sensor and electronics crammed in a 35 mm body. Aside from extreme resolution, it has all the inherent advantages of a MFDB (16-bit color, no AA filter) and disadvantages (very slow, a little glitchy) as well. I know I won't match the look with another DSLR, but hoping to pick up a lot of advantages too.
-Bill
Hi Gandalf
I recently used the 5DmkII and a H3DII on a beauty shoot. I have to say that I was pleasantly surprised at the look of the 5DmkII.
Usually it takes a round of sharpening to bring out the texture of the skin, but this time the image felt more "3 dimensional" right away upon capture.
To put it in context, when shooting with my medium format digital backs I rarely see a need for sharpening the image. Better microcontrast for sure.
With a DSLR sharpening is always necessary.
After the modification the amount of sharpening I use has decreased.
On a pixel level, edges are more distinct
I always felt that the difference between my MF files and DSLR files was that the DSLR image was always abit flatter. I attribute this to the way that pixels on the edges tend to blend into each other abit.
I believe that's why also you see a difference with your leica
Overall I was hoping that the effect would be more prominent but I am aware that the modification removed only one AA filter, (the horizontal one?) whereas the other AA filter (vertical) is stuck onto the sensor itself. Maybe that's why the Leica has a more distinct look?
But for me every single bit of resolution counts in the kind of work I do which is detail intensive, alot of small edges and details. It's abit of a psychological effect too seeing the image straight out of the camera looking nice and crisper to a degree (slightly better microcontrast)
To clarify, I'm speaking from my experience using a 1Ds MkII and a 5D MkII
Hopefully I have more examples to share in the future!
-
Hi,
I don't think that Sony is any better than Canon, Nikon or anything else. I have a Sony Alpha 900, and it is a good camera but I don't see any magic around it.
In order to achieve optimum sharpness we need to put the camera on a steady tripod, adjust autofocus for each lens, use mirror lockup and cable release/self timer and use optimal aperture.
Best regards
Erik
Agreed, although I've seen Andrey Tverdokhleb (RPP designer) mention that the A900 has a relatively weak AA filter. Also, it depends on the subject. With the near-MFDB color separation of the A900, you get more details in the blues than Canon, so, like Jinq q mentioned about the 5Dii, skin details are better with A900....but, we're splitting hairs as usual.
-
Overall I was hoping that the effect would be more prominent but I am aware that the modification removed only one AA filter, (the horizontal one?) whereas the other AA filter (vertical) is stuck onto the sensor itself. Maybe that's why the Leica has a more distinct look?
But for me every single bit of resolution counts in the kind of work I do which is detail intensive, alot of small edges and details. It's abit of a psychological effect too seeing the image straight out of the camera looking nice and crisper to a degree (slightly better microcontrast)
Ahh, very good to know that there is still 1 filter that stays. That is likey why the Kodak SLRc or n, and the leica are very noticably sharper vs this hotrodding. I think you did fine on the test for what you intended to get out of it, but I don't see much of a difference. I don't know if the peppery subject matter helped or hindered the detection?
Thanks for posting, as I think it is informative. I think putting your tests public is a good thing, it can only perfect it. But often there are those that squeel to have superior image..(hehe pun) .
I too would like to see another example. Would also help if you circle the area you did critical focus on.(preferable in the center of the lens.