Ad
Ad
Ad
Pages: [1] 2 3 »   Bottom of Page
Print
Author Topic: Leaf statement: digital sensors & WA photography  (Read 12840 times)
yaya
Guest
« on: October 24, 2006, 11:02:05 AM »
ReplyReply

Hello everyone, the following statement has been published today on several online boards,

Best regards, Yair

Yair Shahar | Leaf EMEA | Regional Manager |
mob: +44 77 8992 8199 | yair.shahar@kodak.com


Cast Effects in Wide Angle Photography

Over the last few months, some image related issues involved in using high resolution Leaf backs with wide-angle, non-retrofocal lenses and camera movements, have been receiving considerable attention on various forums.
We have been thoroughly investigating the issue and would like to share what we have discovered. We want photographers to understand the problem and to appreciate the available solutions.

1. The problem is encountered mainly by our architectural customers, in the form of slowly changing and/or abrupt color cast differences across the image, almost always in cases where non-retrofocal, wide angle lenses are used.

2. Most photographers never encounter this problem. It requires specific optical equipment to recreate it.

3. The problem is a sensor problem. It is not restricted to a particular type of sensor but occurs in different forms in all sensors from high-end 39MP to low-end 3MP sensors. A technical primer about different cast effects and their possible causes can be found at http://www.leaf-photography.com/files/docs...Photography.pdf

4. Leaf and its customers originally expected that the gray uniformity found in the Valeo 22 and the Leaf Aptus 22 would be preserved in the Leaf Aptus 75 and Leaf Aptus 65, despite the considerably smaller size of the pixel (52 sq mm. instead of 81 sq mm.). Gray uniformity is indeed preserved for regular retrofocal lenses where our competitors still suffer from non-uniformity.

5. However, in the wide angle, non-retrofocal domain, the Leaf sensors have not delivered the expected standard of gray uniformity. The problem is not with the abrupt cast change ("centerfold" line) which is a specific anomaly that we expect to solve but with the slowly changing cast effect. After considerable research and discussion, we have regretfully reached the conclusion that with current chip technology, image sensors will probably continue to produce cast for some wide angle lenses and camera movements.

6. The team at Leaf continues to investigate this problem from all aspects - electronic and optical - in search of an effective hardware solution. For now, we have developed and field-tested an initial, practical, software solution which enables the photographer to correct any cast effect, caused either by specialty lenses or by camera movements. The solution is comprehensive but requires some workflow effort. The software can be downloaded from
http://www.leaf-photography.com/pages/supp...n_Adjuster.aspx.

7. We are also working on further software solutions that will reduce the amount of effort required to correct the cast effects. These improvements will be made public as soon as they are ready.
We ask you to please take the time to read the technical bulletin and appreciate the nature of the problem and how it is being dealt with at Leaf. Any suggestions for improving the current solution or for new avenues of investigation are of course most welcome.

Thanks,

Stanley Barkan, Chief Scientist,
Leaf Products, Digital Capture Group
Leaf-Photography.com
Logged
froesner
Jr. Member
**
Offline Offline

Posts: 56


WWW
« Reply #1 on: October 24, 2006, 11:54:21 AM »
ReplyReply

Thank you Yair for posting this statement.

I am one of the unfortunate owners of an Aptus75 which produces the centerfold line and color cast used with my Alpa XY and Schneider Digitar 24 / 35 lenses.

If I understand the statement correctly Leaf can get rid of the centerfold line (most likely by replacing my current back as it happened for other users with the same problem) but will not be capable of solving the cast problem at the hardware side. The only possibility for me would be to use the gain adjuster utility to get the clean files I could expect in return for a considerable investment. But even if this software works I would still have to consider my invest in Leaf, 'cause it would slow down my workflow quiet a bit.

If there is anyone on the forum using a phase one or imacon 39Mpix back together with an Alpa SWA / XY could you please indicate to me whether you guys experience similiar problems. This way I could base my decision to stick with Leaf or switch to another manufacturer on some more facts.

Frank
Logged
mtomalty
Sr. Member
****
Offline Offline

Posts: 536


WWW
« Reply #2 on: October 24, 2006, 12:18:43 PM »
ReplyReply

Quote
If there is anyone on the forum using a phase one or imacon 39Mpix back together with an Alpa SWA / XY could you please indicate to me whether you guys experience similiar problems. This way I could base my decision to stick with Leaf or switch to another manufacturer on some more facts.

Frank,

While I don't own either alternate backs you have mentioned nor have I used the Alpa
I have used both on the Hasselblad ArcBody,which offers some tilt and shift possibilities,
and can assure you that the color cast (usually magenta/green) is present on both and requires
a similar sort of software 'intervention' clean the effect.

It is not particularly difficult but does add one more step in ones workflow. So,in short, if you
are happy with your A75,except for having to deal with colorcast,then you will realize no
advantage,in this regard,by considering another brand.

Mark
Logged
eronald
Sr. Member
****
Online Online

Posts: 4262



« Reply #3 on: October 24, 2006, 01:20:33 PM »
ReplyReply

Quote
The problem is not with the abrupt cast change ("centerfold" line) which is a specific anomaly that we expect to solve but with the slowly changing cast effect. After considerable research and discussion, we have regretfully reached the conclusion that with current chip technology, image sensors will probably continue to produce cast for some wide angle lenses and camera movements.

[a href=\"index.php?act=findpost&pid=81999\"][{POST_SNAPBACK}][/a]

So what is causing the centerfold ? Why is centerfold only seen on Leaf/Sinar ?

Edmund
Logged
rainer_v
Sr. Member
****
Offline Offline

Posts: 1134


WWW
« Reply #4 on: October 24, 2006, 02:05:00 PM »
ReplyReply

"4. Leaf and its customers originally expected that the gray uniformity found in the Valeo 22 and the Leaf Aptus 22 would be preserved in the Leaf Aptus 75 and Leaf Aptus 65, despite the considerably smaller size of the pixel (52 sq mm. instead of 81 sq mm.). Gray uniformity is indeed preserved for regular retrofocal lenses where our competitors still suffer from non-uniformity. "

its somehow brave to name your competitors here...... because after my understanding your DALSA competitors have found solutions which LEAF is still searching .... let me tell you the background for my statement:

what suprises me is that there is exatcly NO statement to the centerfold issue.... of course it's offset is subtracted in the recommended gain calibration software as well as any color cast offsets,- but this helps only to a certain amount, after my understanding this is not the solution to get rid of this issue, if you have a sensor which shows the centefold issue. Many dalsa 33mp sensors do so,- raise up the gradation enough and there will be situations where you might see it ,- not only with shift lenses!

after my experience and after my tests with various backs ( e75 sinar ) this issue is system immanent and can be eliminated only by a clever software solution. in the last months i was involved very closely  in the developement of the programm of the eMotion DNG converter by stephan hess,-  central point of this software was -  beside the conversion to dng - the elimination of the "centerfold" issue. one other of eMotionDngs features is the elimination of color casts nd vignetting by using a calibration shot,- something Leaf does with the separate gain adjuster tool.
although i want to point out that we improved the eMotionDng programm to a tool with a highly automated batch workflow,- which is by far the fastest tool i have seen untill now,- allowing to convert hundreds of e75 and e22 files to dng without beeing involved,- if the source folder once is structured accordingly.

another big step is that stephan wrote a highlight recovery which is very effective till up to 2 stops in the highlights, compared to the original sinar software ( which is comparable to the leaf software in this  ). it allows a far more  relaxt shooting, because you can take much less care regarding blown hilights. eMotionDNG does not cut the information, if at least one color channel still is under 255, and even makes an edjucated guess when all channels exceed 255 -
the problem being to avoid color casts in extreme cases.,  most, probably all other existing mf software goes the most simple way,- doing nothing or blowing all channels up to 255.

i just finished my last shooting. 22 weeks in bangkok,- shooting the new bangkok airport. i started with the sinar e22 and changed after may to the e75,- and asked stephan to help me to make the e75 usable,- which means to deliver absolute neutral files without centerfold issues, and with a fast workflow- i am  using wide angle shift lenses from 28 - 90mm beside h-blad tele lenses with 150+250mm. very important for me was a fast and practical workkflow...... from file numbering,  over batch whiteshot processing to combining shots and respective whiteshot  automatically  when batchconverting the files.. i am excited about the resulting image quality,- its such a great step from the starting point where we have been in May thats its hard to believe that the same back has produced this....
it seems so that our starting point in May is more or less the point where leaf currently is,- so i would say its truly brave to point out the state of your competitors in LEAFs post above.


finally SINAR took over the parts of the software from stephan which bring solutions to the various sensor problems the dalsa 75 brought with it...... and everything`s fine now so far,- at least with the e75.
« Last Edit: October 24, 2006, 02:27:46 PM by rehnniar » Logged

rainer viertlböck
architecture photographer
munich / germany

www.tangential.de
rethmeier
Sr. Member
****
Offline Offline

Posts: 780


WWW
« Reply #5 on: October 24, 2006, 03:10:34 PM »
ReplyReply

Thanks for that explanation Rainer!
IOW all MFDB need some correction for non-retro wide-angles!

My question now is,could the Kodak sensor that used by Imacon and Phase ,be on the same
quality level as the Dalsa sensor that's being used by Leaf?

It's a well known fact that Phase has had their LCC in the Capture One software.

Maybe the P45+ is not such a bad option after-all?

I like the idea of longer exposure times and a higher ISO!

However the e-75 is still No1 on my shopping list.

Regards,
Willem.
Logged

Willem Rethmeier
www.willemrethmeier.com
Sydney Australia
Eric Zepeda
Jr. Member
**
Offline Offline

Posts: 89


WWW
« Reply #6 on: October 24, 2006, 03:15:11 PM »
ReplyReply

maybe Leaf should hire Stephan  

Seriously though Rainer, I'm glad you've found a solution for your workflow. While I'm not thrilled to be using the plexi for Phase Lcc, it sure looks alot better with it than without. As someone stated previously (Mark?) it's another step in the workflow at this point.

Maybe at PhotoPlus all MFDB owners with Lcc issues can all meet under a big white plexi tent and have a beer while wearing special 3D glasses that are magenta in one lens and green in the other...after applying a BC (beer correction) profile everyone should be just fine.
Logged

yaya
Guest
« Reply #7 on: October 24, 2006, 03:56:54 PM »
ReplyReply

Rainer hi,

I am glad that Stepahn's clever tool is now working for you and that the results you are now getting are better than some of the ones shown printed in the booth at Photokina.

Our statement suggests that the current solution for color casts (and the centerfold is indeed an extreme case of cast, due to a certain chip design) is software based, much like Stephan's tool.
It also suggests that we will continue to work and finding a more elegant, effortless solution.

The Leaf Custom Gain Adjuster allows for removal of lens cast (inc. centerfold) and controlled removal of lens falloff, while saving the corrected files as either compressed or uncompressed, retaining all the necessary colour and tonal information, such as ICC input profile that cannot be carried by DNG.

These corrected files can be then taken back into Leaf Capture or any other RAW converter that supports our files for further adjustments/ editing/ processing.

Highlight recovery depends more on what the sensor records at the limit (in 14 or 16 bit) and how it translates to ink on paper in 8 bit (248 on the best RGB printer) then on just "making up" numbers. Your "conclusion" that the Leaf software is "comparable" to CaptureShop in this respect is somewhat mis-leading.

Yair
Logged
rainer_v
Sr. Member
****
Offline Offline

Posts: 1134


WWW
« Reply #8 on: October 24, 2006, 04:09:29 PM »
ReplyReply

Quote
Rainer hi,

I am glad that Stepahn's clever tool is now working for you and that the results you are now getting are better than some of the ones shown printed in the booth at Photokina.

Our statement suggests that the current solution for color casts (and the centerfold is indeed an extreme case of cast, due to a certain chip design) is software based, much like Stephan's tool.
It also suggests that we will continue to work and finding a more elegant, effortless solution.

The Leaf Custom Gain Adjuster allows for removal of lens cast (inc. centerfold) and controlled removal of lens falloff, while saving the corrected files as either compressed or uncompressed, retaining all the necessary colour and tonal information, such as ICC input profile that cannot be carried by DNG.

These corrected files can be then taken back into Leaf Capture or any other RAW converter that supports our files for further adjustments/ editing/ processing.

Highlight recovery depends more on what the sensor records at the limit (in 14 or 16 bit) and how it translates to ink on paper in 8 bit (248 on the best RGB printer) then on just "making up" numbers. Your "conclusion" that the Leaf software is "comparable" to CaptureShop in this respect is somewhat mis-leading.

Yair
[a href=\"index.php?act=findpost&pid=82069\"][{POST_SNAPBACK}][/a]

hi YAYA,
thank you for your flowers regarding my images at photokina,- for sure they would looked much better using an aptus22 not an emotion22..... dont you think so?

( anyway havent seen the prints myself cause i havent had time to be there,- some people told me they liked them,- maybe they also dont know the file quality which would have made your sensor. )

and about highlight recovery: you have not to make so much wind with your explanation,- just shot a file with the e75 and the same with the a75, devellop the raws with the eMotionDNG konverter and with the LEAF software and you will easily  see what about i write,- or this time you could do what you recommended me last time: ask stephan about how he makes it possible that there is detail in his program where the others show  pure and blownout white.
« Last Edit: October 24, 2006, 04:21:32 PM by rehnniar » Logged

rainer viertlböck
architecture photographer
munich / germany

www.tangential.de
yaya
Guest
« Reply #9 on: October 24, 2006, 04:39:20 PM »
ReplyReply

Rainer,

I wasn't refering to the photographic qulaity of those prints since they were very pretty. However they did show some significant purple/ magenta/ green cast, which lead me to believe that they were from the "pre Stephan" period.

I apologize if I sounded offensive.

The test that you are suggesting is something that I do on a regular base with various systems and platforms.

One thing I can say is that no one has ever complained about lack of highlight detail on a Leaf back.

Respectfuly

Yair

Yair Shahar | Leaf EMEA | Regional Manager |
mob: +44 77 8992 8199 | yair.shahar@kodak.com
Logged
narikin
Sr. Member
****
Offline Offline

Posts: 867


« Reply #10 on: October 24, 2006, 04:56:25 PM »
ReplyReply

well after all the beating up of Phase for their occasional LCC issue, I hope those same Leaf owners are now going to enjoy eating their hats over this one.
just tell me - sauteed or broiled with herbs?

I read the white paper and it STILL doesnt properly explain why there is a direct seam down the middle of the Leaf images. it explains why there is an LCC issue with true wide angle lenses, as we all knew, but no real explanation of the centrefold crease.
Logged
yaya
Guest
« Reply #11 on: October 24, 2006, 05:11:53 PM »
ReplyReply

Here is Stephan Hess's
Explanation which in most parts is technically accurate.

Yair
Logged
brumbaer
Jr. Member
**
Offline Offline

Posts: 67


« Reply #12 on: October 24, 2006, 05:24:13 PM »
ReplyReply

Quote
Here is Stephan Hess's
Explanation which in most parts is technically accurate.

Yair
[a href=\"index.php?act=findpost&pid=82088\"][{POST_SNAPBACK}][/a]

I'd for one would be interested in which part it isn't accurate  

Kind regards
Stephan
Logged
Boghb
Guest
« Reply #13 on: October 24, 2006, 05:38:30 PM »
ReplyReply

In my experience, the P45 produces some lens cast with every lens, but the amount varies widely depending on lens quality and design.  

When used with a Schneider 24mm digitar, there were strong and obvious (in the preview) blue/yellow and magenta/green shifts in the center and corners.  This was effectively removed by the LCC tool, but you needed a reference shot for every set of captures.

When used with a Hassy V and CF lenses, there was also noticeable cast with every lens.  Surprisingly, the most cast was produced by the 120 Makroplanar and 250 SA.

I kept my P45 but changed all my lenses.  I now use HC lenses, which produce far less cast, and Rodenstock HRs, which are retrofocus.  There is still some cast in every image, but one-time reference shots fix that.  Use of the LCC tool is also necessary to get the correct white balance.

Phase gave me good support (after an initial lag) and eventually resolved my problem by allowing giving me a new H-back free of charge.  But in my opinion, they were not forthcoming enough with their cutomers in communication and information sharing.  They still won't admit that lens cast is a "problem".

I am impressed with Leaf for acting in such a forthright and honest manner.  This is how you keep your customers even when there is an unexpected problem with your flagship product.
Logged
yaya
Guest
« Reply #14 on: October 24, 2006, 05:38:36 PM »
ReplyReply

Quote
I'd for one would be interested in which part it isn't accurate   

Kind regards
Stephan
[a href=\"index.php?act=findpost&pid=82091\"][{POST_SNAPBACK}][/a]

  You forgot to mention the RGB bayer pattern and the method being used for laying it on the pixels, which also "contributes" to this complexity  

Cheers, Yair
Logged
rainer_v
Sr. Member
****
Offline Offline

Posts: 1134


WWW
« Reply #15 on: October 24, 2006, 06:04:13 PM »
ReplyReply

Quote
Rainer,

I wasn't refering to the photographic qulaity of those prints since they were very pretty. However they did show some significant purple/ magenta/ green cast, which lead me to believe that they were from the "pre Stephan" period.

I apologize if I sounded offensive.



Yair Shahar | Leaf EMEA | Regional Manager |
mob: +44 77 8992 8199 | yair.shahar@kodak.com
[a href=\"index.php?act=findpost&pid=82082\"][{POST_SNAPBACK}][/a]

Hi Yair,
thank you for clarifying, maybe i missunderstood you......
just for interest...: which of the images show this  "significant purple/ magenta/ green cast" ?
here they are:
[attachment=1093:attachment]
« Last Edit: October 24, 2006, 06:06:08 PM by rehnniar » Logged

rainer viertlböck
architecture photographer
munich / germany

www.tangential.de
ericstaud
Sr. Member
****
Offline Offline

Posts: 384


WWW
« Reply #16 on: October 24, 2006, 06:30:05 PM »
ReplyReply

"which of the images show this "significant purple/ magenta/ green cast" ?"

# 1, 3, 4, and 6 show color casts.

-Just to be clear, the cast is not "significant", but present.  I am also judging jpegs on a web browser.  The pictures are amazing.  I would love to have that opportunity.  And in each case, the cast does not detract from the overall impact of the images.  It is likely only a very anal architectural shooter would notice.
« Last Edit: October 24, 2006, 06:39:17 PM by ericstaud » Logged
rainer_v
Sr. Member
****
Offline Offline

Posts: 1134


WWW
« Reply #17 on: October 24, 2006, 06:38:26 PM »
ReplyReply

Quote
"which of the images show this "significant purple/ magenta/ green cast" ?"

# 1, 3, 4, and 6 show color casts.
[a href=\"index.php?act=findpost&pid=82102\"][{POST_SNAPBACK}][/a]

color casts which are caused by any sensor shifts? i hardly doubt that.....    more because all images are taken with the 22mp dalsa chip,- which has not suffered from the actual problems.
maybe you dont like my way of editing the images,- but sensor-caused colorcasts?
i would simply deny that.
the slight shift at the edges of image 1and6 is made by the 28HR, i  think every lense with a fov as that shows some blueish-yellow cast,- at least an engeneer from schneider-kreuznach explained that to me some years ago, cause i phoned them to discuss the blueish - yellow casts which show the 4x5" wides from them ( all under 90mm,- it becomes very visible with the 57+47+37xl lenses ). its an optical law, no way to escape it,- although inverting, as phase always suggested with lcc and now also the dalsa sensors need with whiteref. shots inverts that casts also. but that has nothing to do with uneven sensor reproduction. these shots are not "treated" with whiterefernces.
« Last Edit: October 24, 2006, 06:48:25 PM by rehnniar » Logged

rainer viertlböck
architecture photographer
munich / germany

www.tangential.de
brumbaer
Jr. Member
**
Offline Offline

Posts: 67


« Reply #18 on: October 24, 2006, 06:43:41 PM »
ReplyReply

@yair

Sorry,
you implied that I wrote something which is not accurate, which implies I wrote something that is wrong.

Now you say I left something out, which is quite different.

By the way I didn't leave it out, I said "This has to do with the manufacturing process. Whatever the reason for having those areas is doesn't matter, important is that they exist."

This does not reduce the accuracy of my statement.
If anything it increases it, because the only information I have about the RGB Pattern thingie is from you. And I will not quote things I can't verify in a post in which I want to give information and not opinion.

Talking about opinions, here is one, so a big IMHO:

I've got the feeling your reaction to anybody who is of a different opinion is to try to discredit him, especially in a difficult situation as the one your in.
Your offhand "which in most parts is technically accurate" implying wrong information, without being able to proof it, is one such case.
Or trying to tell people that the images of a photographer are bad and when somebody asks, they are not bad, they suddenly have a cast. And when you look at the images there is no cast.

I've read some helpful and interesting posts from you, so I hope this all is a misunderstanding or a symptom of stress or probably it is just the so called language barrier.

@all
I apologize for wasting bandwidth on a rather personal matter.
But after having read the discussion about anonymity and the "higher worth" of information of people with "names" I think that especially members with "names", should be careful with their posts and should be aware of their responsibilty as opinion makers, and they should be told, if they seem to forget.


Kind regards
Stephan
Logged
ericstaud
Sr. Member
****
Offline Offline

Posts: 384


WWW
« Reply #19 on: October 24, 2006, 06:45:26 PM »
ReplyReply

Hi Rainer,

I was editing my previous post to be a little more complete while you were replying at the same time.  My Dalsa 22mp chip had color casts.  Always subtle, always hard to spot.  I often find that with the Dalsa 75 chip that I do not see the color casts until I have run the Custom Gain Adjuster and compare the images back and forth in the preview window of Bridge.  Please don't take this as any kind of criticism.  I feel most of my photos shot with the 22 sensor have a little cast here or there which no one else notices.
Logged
Pages: [1] 2 3 »   Top of Page
Print
Jump to:  

Ad
Ad
Ad