Pages: [1] 2   Go Down

Author Topic: accuracy of "focus and recompose"?  (Read 9789 times)

photolinia

  • Jr. Member
  • **
  • Offline Offline
  • Posts: 78
accuracy of "focus and recompose"?
« on: May 18, 2009, 05:01:05 pm »

Hi everyone,

Once again, forgive me for bringing up a topic that I know was already discussed...

Coming from Nikon's excellent 50+ point AF system that is lightning fast and accurate, I've been spoiled into pretty much always relying on
AF and never doing any manual focusing.

Getting into the H3D - a single point, and much slower AF, I talked to several experienced users and typically was told that
getting the hang and learning how to focus manually is the way to go and that "focus and recompose" will not be as accurate - especially with a single point
AF and if my focus point is close to the frame edge...

I agree that MF is the way to go - especially in a studio, with a still model, with the camera on a tripod - tethered to a computer + checking focus on the computer screen, but what about doing this on location without a laptop and when your models are not perfectly still?

I spent all day yesterday shooting outside - using manual focus exclusively and my results are very much all over the place.
I was using the 100mm f/2.2 lens and was able to nail the focus on only about 15% of the shots.  The rest are anywhere from slightly soft to completely
out of focus.  Problem is - I just do not trust my eyes enough after being spoiled by Nikon...

I'm curious as to how many people do the "focus/recompose" thing and find it accurate even at f2.2.
Should I save manual focus for studio only, or should I keep trying to master the MF in the field?

Thanks!
-ilya
Logged

tho_mas

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1799
accuracy of "focus and recompose"?
« Reply #1 on: May 18, 2009, 05:23:19 pm »

Quote from: photolinia
or should I keep trying to master the MF in the field?
I'd say you should.
Here's a simplified explanation: http://www.mhohner.de/recompose.php
Though the focus plane (blue lines) is not strictly straight (IMHO) the outline shows the principle.
Logged

photolinia

  • Jr. Member
  • **
  • Offline Offline
  • Posts: 78
accuracy of "focus and recompose"?
« Reply #2 on: May 18, 2009, 09:10:29 pm »

I guess my biggest question is what does the plane of focus really look like?
Is it really flat - like the blue lines in the link?
or is it more round - like the broken line?  
I would think that it's more rounded rather then flat, but I'm not sure...

-ilya
Logged

Kitty

  • Full Member
  • ***
  • Offline Offline
  • Posts: 126
    • http://
accuracy of "focus and recompose"?
« Reply #3 on: May 18, 2009, 11:50:09 pm »

Quote from: photolinia
I guess my biggest question is what does the plane of focus really look like?
Is it really flat - like the blue lines in the link?
or is it more round - like the broken line?  
I would think that it's more rounded rather then flat, but I'm not sure...

-ilya

The plane of focus is the blue parallel line not broken radius line.
I did the test on Leica M8 to prove it.
And it is true!
Someone recommend to move camera in parallel to recompose instead of turning camera.
It helps.

kitty
Logged

Dustbak

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2442
    • Pepperanddust
accuracy of "focus and recompose"?
« Reply #4 on: May 19, 2009, 01:47:29 am »

I know about the theory of why you should not focus/recompose. On my H2F I focus/recompose most of the time. I most cases focus ends up where it needs to be.

When I work from tripod with static subjects I use manual focus.

When I have serious doubts focus/recompose is not going to work I use MF but in most cases focus/recompose generates a higher hit rate.

When upclose and personal I use MF.

I do keep in mind the way my focal plane is formed, maybe, I am not sure I subconsciously correct for that while shooting. I don't really pay attention to that.

Focus with the 39MP system is truly critical. Miss it by the smallest of margins and it will really look like crap.
« Last Edit: May 19, 2009, 04:19:07 pm by Dustbak »
Logged

pegelli

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1664
    • http://pegelli.smugmug.com/
accuracy of "focus and recompose"?
« Reply #5 on: May 19, 2009, 06:22:54 am »

Quote from: Kitty
I did the test on Leica M8 to prove it.
And it is true!

I think it's more a function of the lens rather than the camera, however most lenses (and certainly the high quality Leica ones) have a flat plane (the blue line) as their prime focus. There's some WA zooms that show some field curvature (eg. the Tamron 17-50/2.8 is known for that) however it's still far away from being the inside of a sphere.

On the other hand I think only for large apertures and large angle swings it will start making a real difference. For most other cases the very small difference will fall within the dof of the lens anyway. So for instance with portraits with a 85/1.4 I'm very careful not to recompose after locking focus on the eyes, but in other situations I have used the technique quite successful.
Logged
pieter, aka pegelli

jimgolden

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 410
    • http://
accuracy of "focus and recompose"?
« Reply #6 on: May 19, 2009, 12:34:58 pm »

I'm shocked focus and recompose would work on a MF system @ f/2.2 or 35mm @ 1.4...I wouldn't even dare...
Logged

photolinia

  • Jr. Member
  • **
  • Offline Offline
  • Posts: 78
accuracy of "focus and recompose"?
« Reply #7 on: May 19, 2009, 12:54:26 pm »

I'm going to experiment with this, but I have a feeling that right now AF & recompose will still do better than MF for me...
doing MF is very stressfull right now for me   I sweat so much over the focusing, I hardly have time to concentrate on other things like composition and background...
Logged

julius0377

  • Jr. Member
  • **
  • Offline Offline
  • Posts: 56
accuracy of "focus and recompose"?
« Reply #8 on: May 19, 2009, 02:44:47 pm »

Quote from: photolinia
I spent all day yesterday shooting outside - using manual focus exclusively and my results are very much all over the place.
I was using the 100mm f/2.2 lens and was able to nail the focus on only about 15% of the shots.  The rest are anywhere from slightly soft to completely out of focus.

It was just like this for me in the beginning going from solely shooting canon to a mixed canon, nikon and hblad environment. Lots of missed focus, in particular outside the studio.

It gets a lot better. If I had around a 20-30 percent successrate in the beginning, after a couple of months I was up to around 70-80 percent success, and now some years later I'm rarely seeing one or two images out of a hundred that are truly soft.

I have some techniques that have settled in my body after using the blad. I move my head/body back when recomposing (I don't know how much, it depends on how far the recompose is and how far away the subject is... sort of gets "in your blood").

Sometimes I focus and recompose, but move the focusing ring a bit to get eyes and such so snap into focus, and the amount I move the ring has sort of become habit like a focus puller in the movies know the exact speed to turn the ring depending on the speed of the dolly/wagon.

Keep on using the camera, but make the pace slower at first, after a while you will be speedy and efficient with a very dependable success rate. Try to use apertures in the 8 to 16 range when on strobes unless a larger opening is neccessary for artistic or DOF purposes.
Logged

Wayne Fox

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 4237
    • waynefox.com
accuracy of "focus and recompose"?
« Reply #9 on: May 19, 2009, 02:48:15 pm »

Quote from: photolinia
I'm going to experiment with this, but I have a feeling that right now AF & recompose will still do better than MF for me...
doing MF is very stressfull right now for me   I sweat so much over the focusing, I hardly have time to concentrate on other things like composition and background...

Focus and recompose is not new to auto focus cameras.  Most of us relied on viewfinders with various enhancements, such as split image focusing screens and focus and recomposing was typical as well.

In the example given the case of focus and recompose is exaggerated because it is focusing on an object that appears to be fairly close to the camera, using a pretty wide angle view, and an object that is at the extreme edge of the image.  I'm not sure I've ever been concerned about the sharpness of something that far to the edge of my image.  Certainly there are cases where objects close to the edge need to be in focus, but generally it's because the subject covers the entire field, such as a family group, so I'm still focusing on something towards the center. Generally when focusing then recomposing you are still focusing on something that is much closer to the center of your frame.  If using a telephoto lens, the angles also don't add up to make much difference, and even with normal to wide lenses the subject may be far enough away and close enough to the center as to not be problematic.

It is good to be aware of the possible issues, but manually focusing a mfdb is very problematic ... auto focus and recompose may be more accurate most of the time.  In circumstances where recomposing may compromise image sharpness, hopefully you are using a tripod and can verify your manual focusing by zooming in on the preview.

If we could only get a form of Live view just for focusing
Logged

Frank Doorhof

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1522
    • http://
accuracy of "focus and recompose"?
« Reply #10 on: May 19, 2009, 02:55:26 pm »

I've been using center focus point only, ever since I took up a camera (when I got AF camera's )
There is alot of talk about not being able to recompose and shoot but I always have the focus exactly where I want (well ok let's say 99%).

With the AFD/III there is still only a center focus point and even with the RZ67ProII I often use the center point to focus and recompose on the studio stand.
It's probarbly indeed getting used to it.

I have to add shooting wide open on a 39mp camera is asking for problems with a 2.2 lens.
I own the 1.9 80mm and when I use that one wide open it's a wonderful lens but getting focus accurate is alot of practice, I can expect that when you switch from a DSLR to MF you will have to work for a few weeks before you will master that.

Remember that the DOF is MUCH less than with a FF DSLR.
Logged

Toto

  • Newbie
  • *
  • Offline Offline
  • Posts: 26
accuracy of "focus and recompose"?
« Reply #11 on: May 19, 2009, 03:41:17 pm »

Hello

Coming from Nikon world to H3DII, the main problem of focus I've had is when I do not recompose. I shoot a portrait outside, and  the system is focusing on the backdrop. Really, difficult to spoil this with Nikon, but with Hassy, yes. I've to think about it each time, not to rely too much on AF !
Logged

paratom

  • Full Member
  • ***
  • Offline Offline
  • Posts: 205
accuracy of "focus and recompose"?
« Reply #12 on: May 20, 2009, 03:15:24 am »

Quote from: Toto
Hello

Coming from Nikon world to H3DII, the main problem of focus I've had is when I do not recompose. I shoot a portrait outside, and  the system is focusing on the backdrop. Really, difficult to spoil this with Nikon, but with Hassy, yes. I've to think about it each time, not to rely too much on AF !

Also someone using Nikon and MF (Hy6).
My problem is that with the Hy6 I cant see how big the AF-sensor is and where its actually located plus I feel it is "too" big.
With the Nikon I can set place the AF sensor "on the eye" of the portrait, with the Hy6 I just can focus on the face.
My keepter rate regarding 100% accurate focus is quite lower with the MediumFormat system.
Manual Focus though works better compared to smaller cameras because of the nice and bright viewfinder.
Right now my keeper rate with correct focus for subjects not beeing still (portrait, kid, etc.) when shooting the lenses open is not really that encouraging.
However I still find it worth working on it because I like the files better than those from the d3x.
Logged

Morgan_Moore

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2356
    • sammorganmoore.com
accuracy of "focus and recompose"?
« Reply #13 on: May 20, 2009, 03:42:32 am »

I have been saying probably for three years that the limit of resolution is the human handling the camera

The D3, the focus is not spaced enough

The H1 Centre is of course a joke

On the blad I focus bracket (rock my body mildly forwards and backwards handheld) and bank on the getting a low percent but some cracking images, on the nikon my shots are all 'OK' using generally misplaced focus point or MP losing compromised composition

Depentdant on the client I will choose the tool

I am convinced that the future of critical focus above the 20mp resolution will require quality zoomable liveview (preferably with clickable AF area)

check out some compacts - they rock with face recognition for $250

S
« Last Edit: May 20, 2009, 03:43:41 am by Morgan_Moore »
Logged
Sam Morgan Moore Bristol UK

pegelli

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1664
    • http://pegelli.smugmug.com/
accuracy of "focus and recompose"?
« Reply #14 on: May 20, 2009, 04:03:43 am »

Quote from: Wayne Fox
In the example given the case of focus and recompose is exaggerated because it is focusing on an object that appears to be fairly close to the camera, using a pretty wide angle view, and an object that is at the extreme edge of the image.  I'm not sure I've ever been concerned about the sharpness of something that far to the edge of my image.

Wayne, the example drawing is quite extreme, but in real life even with longer lenses and a smaller arc there can be an issue.

Following example:

85/1.4 lens taking a portrait wide open, focus needs to be nailed on the eyes (at least according to conventional wisdom)

Assume 1.5 meter distance, full frame camera, only using center spot focus point. Dof under these conditions is +/- 1.3 cm

Focus on the eyes and then change the composition by turning the camera such that the center spot sensor moves about 20 cm down on the face, such that the eyes end up in the top 1/3 of the frame. This is a rotation of less than 10 degrees.

distance change to the eyes by turning the camera is sqrt((150)^2 + (20)^2) - 150 = 1.3 cm (by coincidence the same number as the dof)

This means that the eyes are no longer in the middle of the dof zone, but right at the edge, so the slightest focussing error to the wrong side will render the eyes not as the sharpest point in the frame.

Add to that the risk of moving slighly forward/backward during recomposing and the chance of focus errors start stacking up. So therefore under these circumstances using an AF sensor very close to the point where you want the peak sharpness and avoid recomposing still makes a lot of sense.
Logged
pieter, aka pegelli

Jim Pascoe

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1131
    • http://www.jimpascoe.co.uk
accuracy of "focus and recompose"?
« Reply #15 on: May 20, 2009, 06:30:34 am »

I have read about this topic before and I have to admit was a bit sceptical about the whole thing.  However I do shoot at wide apertures mostly, and am nomally photographing people.  My technique has always been to use the central point and to focus and re-compose.  When close up the eyes are usually the thing to get really sharp.  With the 50mm 1.4 in particular, the focus has often been off using this technique, and sometimes I have blamed the lens!

This morning, having read this post I went out into the garden with my 135 and 50mm lenses, both set to f2.  Well the upshot is that I have learned an important lesson.  At wide apertures, focus and re-compose is not good enough, at least not if you are being critical.  I have included two sets of pictures to illustrate my test, one taken with the 50mm and the other with the 135mm lens.  For interest they were mounted on my Canon 1DS mk3, and were hand-held, seeing as this is how I normally work.  I did in fact do about ten different set-ups, with the same result in all cases.

Each set shows the full image, then the other two pictures show the centre point selected and re-compose, followed by the same picture but using the focus point over the subject.
The shot of the flowers was focussed on the bloom at the top.

This is my first attempt at uploading pictures to this forum, so I hope they are ok. Also, I know this is in the MFDB section, but I suppose the principle is the same!

Jim
[attachment=13806:135mm_f2.jpg] [attachment=13807:centre_f...s_point1.jpg][attachment=13808:off_cetr...s_point1.
jpg]
[attachment=13809:50mm_f2.jpg][attachment=13810:centre_f...s_point2.jpg] [attachment=13811:off_cent...s_point2.jpg]

PS.  I did name the files.  Perhaps someone could explain how I get the file names to show on the image.
« Last Edit: May 20, 2009, 06:35:19 am by Jim Pascoe »
Logged

pegelli

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1664
    • http://pegelli.smugmug.com/
accuracy of "focus and recompose"?
« Reply #16 on: May 20, 2009, 06:37:53 am »

Jim, Nice test and thanks for taking the time and posting the results.

My calculation above was on a theoretical basis and it's so much better when backed up by real world results !
Logged
pieter, aka pegelli

Jonathan Wienke

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 5829
    • http://visual-vacations.com/
Logged

R_Medvid

  • Jr. Member
  • **
  • Offline Offline
  • Posts: 87
    • http://kadru.net
accuracy of "focus and recompose"?
« Reply #18 on: May 20, 2009, 01:54:28 pm »

Quote from: Jim Pascoe
This morning, having read this post I went out into the garden ...........

Jim

Hi Jim,

Thanks for your posting. It's not often that you have a quick real-world example of the topic discussed here.

However, looking at the flower out of focus, I am thinking if it being out of the center of the frame (out of the sweet spot) and closer to the edge of the frame (where MTF characteristics worsen especially at the open aperture) could have contributed to the blurr-ness of the flowers.

Just a thought.

THNX
Logged
Roman Medvid
Mamiya AFD / P40+
http://kadru.net

Wayne Fox

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 4237
    • waynefox.com
accuracy of "focus and recompose"?
« Reply #19 on: May 20, 2009, 06:18:58 pm »

Quote from: pegelli
Wayne, the example drawing is quite extreme, but in real life even with longer lenses and a smaller arc there can be an issue.

Following example:

85/1.4 lens taking a portrait wide open, focus needs to be nailed on the eyes (at least according to conventional wisdom)

Assume 1.5 meter distance, full frame camera, only using center spot focus point. Dof under these conditions is +/- 1.3 cm

Focus on the eyes and then change the composition by turning the camera such that the center spot sensor moves about 20 cm down on the face, such that the eyes end up in the top 1/3 of the frame. This is a rotation of less than 10 degrees.

distance change to the eyes by turning the camera is sqrt((150)^2 + (20)^2) - 150 = 1.3 cm (by coincidence the same number as the dof)

This means that the eyes are no longer in the middle of the dof zone, but right at the edge, so the slightest focussing error to the wrong side will render the eyes not as the sharpest point in the frame.

Add to that the risk of moving slighly forward/backward during recomposing and the chance of focus errors start stacking up. So therefore under these circumstances using an AF sensor very close to the point where you want the peak sharpness and avoid recomposing still makes a lot of sense.

Of course, you are using another extreme ... a 1.4 lens at wide open so almost 0 depth of field, and 85mm barely even qualifies as a telephoto for a FF dSLR, let a lone a mfdb.  Granted this is a nice combo for some portrait work (although I prefer using 150-300mm for portraits ... would only use the 85 for groups).

I won't question your math, but from a practicality point of view I would bet that with the hassleblad my chances of getting the images more in focus by trying to manually focus on the eyes  would probably not result in a very successful attempt.

I'd be better off backing up a few feet, allowing some room to crop from that great big file and leaving the eyes dead center, creating my composition later.  Of course I also wouldn't be working at 1.4, since I'm not sure there is a lens that is wider than f/4, which changes tgeh math quite a bit.

I guess my point is while it is something to be aware of, there is no practical and useful remedy for those shooting mfdb (which is what the OP was discussing and having issues with), and most of the time focus and recompose works quite well.  If you are aware of when it might be a concern, then you can take steps to avoid it.
« Last Edit: May 20, 2009, 06:23:00 pm by Wayne Fox »
Logged
Pages: [1] 2   Go Up