E-M5 ACR Baseline Exposure, UniWB, and ETTR

Started May 2, 2012 | Discussions
micksh6
Senior MemberPosts: 2,331Gear list
Like?
Re: E-M5 ACR Baseline Exposure, UniWB, and ETTR
In reply to micksh6, May 3, 2012

gollywop wrote:

The WB doesn't come into play until after the shot is taken and the raw data are processed in-camera by the jpeg engine.

Just noticed this phrase. This is the error in logic. We just recently discussed green live view with UniWB, right? Live view works before the shot is taken, it uses JPEG engine and WB is a part of it.

On the other hand, for performance reasons, it makes sense to use single settings for LV and CDAF algorithm.

 micksh6's gear list:micksh6's gear list
Olympus PEN E-PL5 Olympus E-M1 Olympus M.Zuiko Digital ED 9-18mm 1:4.0-5.6 Panasonic Leica Summilux DG 25mm F1.4 Olympus M.Zuiko Digital ED 40-150mm 1:4-5.6 R +6 more
Reply   Reply with quote   Complain
gollywop
Veteran MemberPosts: 5,304
Like?
Re: E-M5 ACR Baseline Exposure, UniWB, and ETTR
In reply to micksh6, May 3, 2012

micksh6 wrote:

gollywop wrote:

The WB doesn't come into play until after the shot is taken and the raw data are processed in-camera by the jpeg engine.

Just noticed this phrase. This is the error in logic. We just recently discussed green live view with UniWB, right? Live view works before the shot is taken, it uses JPEG engine and WB is a part of it.

There is no question but what the WB is applied to the LV image. But that does not mean that it is part of the cdaf algorithm, which, as I understand it, reads from a number spots directly off the sensor. So it is not shown there is any error in logic. The error could be (and I suspect is) in the synthesis "and WB is a part of it."

Indeed, I am still not convinced the jpeg engine plays any part in the cdaf at all. I would appreciate anyone who knows how it works to chime in. I certainly do not claim to know for sure, but I wouldn't be the slightest bit surprised if there were dedicated circuitry to assess the cdaf, quite apart from the jpeg engine and having nothing directly to do with the full image. There is nothing about the existence of LV that precludes this possibility.

And you were absolutely correct in your earlier post when you suggested the likelihood that I would not find the citations you gave to provide significant evidence.

As a matter of interest, I have just done some tests with my E-M5 with s-af, and there is absolutely no difference in the focusing action or speed with AutoWB, UniWB, muted, vivid, high/low contrast. I haven't messed around with c-af. I'll do that when I get some time.

-- hide signature --

gollywop

-----------

Reply   Reply with quote   Complain
rrr_hhh
Veteran MemberPosts: 5,894Gear list
Like?
Re: E-M5 ACR Baseline Exposure, UniWB, and ETTR
In reply to Detail Man, May 3, 2012

Detail Man wrote:

Can anybody cite any sort of tangible, meaningful, and documented hard evidence (outside of purely anecdotal "impressions") that such is in fact the case for any particular camera model ?/

Yes, Don Parrot tested continuous focus with his running dogs extensively hen he got his E-m5. Make a search. Brian and several others tested it on the E-P3 and confirmed it was the case.

-- hide signature --

rrr_hhh

 rrr_hhh's gear list:rrr_hhh's gear list
Olympus PEN E-P3 Olympus OM-D E-M5 Canon EOS 6D Canon EF 50mm f/1.4 USM Canon EF 16-35mm f/2.8L USM +18 more
Reply   Reply with quote   Complain
rrr_hhh
Veteran MemberPosts: 5,894Gear list
Like?
Re: E-M5 ACR Baseline Exposure, UniWB, and ETTR
In reply to gollywop, May 3, 2012

gollywop wrote:

As a matter of interest, I have just done some tests with my E-M5 with s-af, and there is absolutely no difference in the focusing action or speed with AutoWB, UniWB, muted, vivid, high/low contrast. I haven't messed around with c-af. I'll do that when I get some time.

Geeezzzz you accuse him of not providing evidence, but how did you do your SAF tests ? The AF has reached such a speed that it becomes difficult to measure any difference in SAF. Those who tested tested it with CAF, where it is easy to shoot several sequences and count the number of hits and misses to compare. Their remarks concerns mainly CAF at 4 fps. From there on, it may be assumed that used in less contrasty environments it may results in less focus hunting lenses, but this hasn't been as clearly established as the CAF effects.

That something can't be scientifically explained doesn't prevent the fact that it can be reproduced in experimental environments. Same here those who tested have no explanations as to how it works, but they have results to see it.

-- hide signature --

rrr_hhh

 rrr_hhh's gear list:rrr_hhh's gear list
Olympus PEN E-P3 Olympus OM-D E-M5 Canon EOS 6D Canon EF 50mm f/1.4 USM Canon EF 16-35mm f/2.8L USM +18 more
Reply   Reply with quote   Complain
micksh6
Senior MemberPosts: 2,331Gear list
Like?
Re: E-M5 ACR Baseline Exposure, UniWB, and ETTR
In reply to gollywop, May 3, 2012

gollywop wrote:

Indeed, I am still not convinced the jpeg engine plays any part in the cdaf at all. I would appreciate anyone who knows how it works to chime in.

Nobody knows exactly how it works so it's unlikely that someone will tell you. You won't find any official documents. Manufacturers will never disclose how their cameras work in details. Reverse engineering is sometimes the way to find out.

As a matter of interest, I have just done some tests with my E-M5 with s-af, and there is absolutely no difference in the focusing action or speed with AutoWB, UniWB, muted, vivid, high/low contrast. I haven't messed around with c-af.

You normally won't notice S-AF autofocus difference unless you focus in really low light. With decent light AF speed gain is measured in tens of milliseconds, it's beyond human abilities no notice it. C-AF will show the difference. S-AF in normal light will show too, when you shoot fast moving subjects, but only statistically, when you look at percentage of sharp photos.

I pretty much explained how it most likely works from point of view of image processing software engineer. You may believe it or not, I don't have to convince you. I am all ears if someone provides better explanation.

http://forums.dpreview.com/forums/read.asp?forum=1041&message=41281419
http://forums.dpreview.com/forums/read.asp?forum=1041&message=41302352
http://forums.dpreview.com/forums/read.asp?forum=1041&message=41303014
http://forums.dpreview.com/forums/read.asp?forum=1041&message=41320917

I certainly do not claim to know for sure, but I wouldn't be the slightest bit surprised if there were dedicated circuitry to assess the cdaf, quite apart from the jpeg engine and having nothing directly to do with the full image. There is nothing about the existence of LV that precludes this possibility.

Yes, there is. It takes at least 4096x3 array of coefficients to load into DSP in order to change picture settings from live view to CDAF optimized. Doing that may easily break AF speed requirement of 120 fps rate on some hardware.

 micksh6's gear list:micksh6's gear list
Olympus PEN E-PL5 Olympus E-M1 Olympus M.Zuiko Digital ED 9-18mm 1:4.0-5.6 Panasonic Leica Summilux DG 25mm F1.4 Olympus M.Zuiko Digital ED 40-150mm 1:4-5.6 R +6 more
Reply   Reply with quote   Complain
Iliah Borg
Forum ProPosts: 15,679
Like?
Re: E-M5 ACR Baseline Exposure, UniWB, and ETTR
In reply to micksh6, May 3, 2012

Another reason is that muted ETTR colors and low JPEG contrast impair focusing speed

Try b/w

Reply   Reply with quote   Complain
Detail Man
Forum ProPosts: 14,837
Like?
Re: E-M5 ACR Baseline Exposure, UniWB, and ETTR
In reply to micksh6, May 3, 2012

micksh6 wrote:

I pretty much explained how it most likely works from point of view of image processing software engineer.

What exactly did your direct experience as an "image-processing software engineer" involve ? ..

Reply   Reply with quote   Complain
micksh6
Senior MemberPosts: 2,331Gear list
Like?
Re: E-M5 ACR Baseline Exposure, UniWB, and ETTR
In reply to gollywop, May 3, 2012

gollywop wrote:

micksh6 wrote:

gollywop wrote:

The WB doesn't come into play until after the shot is taken and the raw data are processed in-camera by the jpeg engine.

Just noticed this phrase. This is the error in logic. We just recently discussed green live view with UniWB, right? Live view works before the shot is taken, it uses JPEG engine and WB is a part of it.

There is no question but what the WB is applied to the LV image. But that does not mean that it is part of the cdaf algorithm, which, as I understand it, reads from a number spots directly off the sensor. So it is not shown there is any error in logic. The error could be (and I suspect is) in the synthesis "and WB is a part of it."

Indeed, I am still not convinced the jpeg engine plays any part in the cdaf at all. I would appreciate anyone who knows how it works to chime in. I certainly do not claim to know for sure, but I wouldn't be the slightest bit surprised if there were dedicated circuitry to assess the cdaf, quite apart from the jpeg engine and having nothing directly to do with the full image. There is nothing about the existence of LV that precludes this possibility.

And you were absolutely correct in your earlier post when you suggested the likelihood that I would not find the citations you gave to provide significant evidence.

As a matter of interest, I have just done some tests with my E-M5 with s-af, and there is absolutely no difference in the focusing action or speed with AutoWB, UniWB, muted, vivid, high/low contrast. I haven't messed around with c-af. I'll do that when I get some time.

CDAF is is certainly different topic from UniWB so I created another thread
http://forums.dpreview.com/forums/read.asp?forum=1041&message=41406241

And, BTW, thank you for your post regarding UniWB on E-M5. I will certainly use UniWB when I get this camera. For scene calibration, at least. Form what I see it may be harder to calibrate vs E-P3 but I also hope exposure won't have to be that precise because of extended dynamic range of E-M5. I guess I may be able to afford conservative settings and underexpose by 0.7 EV in low dynamic range cases.

PS. Measuring EV difference between UniWB and AutoWB is less work than shooting grey card to determine WB. And it also doesn't require carrying grey card.

 micksh6's gear list:micksh6's gear list
Olympus PEN E-PL5 Olympus E-M1 Olympus M.Zuiko Digital ED 9-18mm 1:4.0-5.6 Panasonic Leica Summilux DG 25mm F1.4 Olympus M.Zuiko Digital ED 40-150mm 1:4-5.6 R +6 more
Reply   Reply with quote   Complain
gollywop
Veteran MemberPosts: 5,304
Like?
Re: E-M5 ACR Baseline Exposure, UniWB, and ETTR
In reply to rrr_hhh, May 3, 2012

rrr_hhh wrote:

gollywop wrote:

As a matter of interest, I have just done some tests with my E-M5 with s-af, and there is absolutely no difference in the focusing action or speed with AutoWB, UniWB, muted, vivid, high/low contrast. I haven't messed around with c-af. I'll do that when I get some time.

Geeezzzz you accuse him of not providing evidence, but how did you do your SAF tests ? The AF has reached such a speed that it becomes difficult to measure any difference in SAF.

I used a lens that makes the cdaf work in a measurable time. The E-M5 is very fast with the mFT lenses, but not with the standard FT lenses. With my 8-19 or 70-300, there is systematic hunting: a confirmation long, a confirmation short, and a smack right on. The length and nature of that rhythm is very discernible and memorable, and it doesn't change with any of those alterations.

Unfortunately, I can't use the same procedure for c-af, because the E-M5 doesn't do c-af with those lenses, only with the mFT lenses.

That something can't be scientifically explained doesn't prevent the fact that it can be reproduced in experimental environments.

The issue can indeed be scientifically explained. The problem is we don't have the technical knowledge to do so. So at this point we indeed must rely on empiricism. That's fine. Don Parrot's tests, which, by the way, are hardly "extensive," certainly open the issue for serious question.

-- hide signature --

gollywop

-----------

Reply   Reply with quote   Complain
gollywop
Veteran MemberPosts: 5,304
Like?
Re: E-M5 ACR Baseline Exposure, UniWB, and ETTR
In reply to Iliah Borg, May 3, 2012

Iliah Borg wrote:

Another reason is that muted ETTR colors and low JPEG contrast impair focusing speed

Try b/w

Excellent idea.

-- hide signature --

gollywop

-----------

Reply   Reply with quote   Complain
Bob Tullis
Forum ProPosts: 25,280Gear list
Like?
Configure UniWB for Dummies
In reply to gollywop, May 3, 2012

Is there such a writeup?

-- hide signature --

...Bob, NYC
http://www.bobtullis.com

"Well, sometimes the magic works. . . Sometimes, it doesn't." - Little Big Man
.

 Bob Tullis's gear list:Bob Tullis's gear list
Olympus E-M1 Rokinon 7.5mm 1:3.5 UMC Fisheye CS Voigtlander Nokton 42.5mm F0.95 Voigtlander Nokton 17.5mm F0.95 Aspherical Olympus 12-40mm F2.8 +17 more
Reply   Reply with quote   Complain
dcassat
Senior MemberPosts: 1,141Gear list
Like?
Re: Configure UniWB for Dummies
In reply to Bob Tullis, May 3, 2012

+1 Bob.

Golly you did the step by step right? As I remember there were a lot of steps...?
--
Dan

 dcassat's gear list:dcassat's gear list
Sigma DP1 Merrill Olympus OM-D E-M5 Olympus M.Zuiko Digital ED 9-18mm 1:4.0-5.6 Panasonic Lumix G 20mm F1.7 ASPH Panasonic Leica D Summilux Asph 25mm F1.4 +5 more
Reply   Reply with quote   Complain
Detail Man
Forum ProPosts: 14,837
Like?
Re: E-M5 ACR Baseline Exposure, UniWB, and ETTR
In reply to gollywop, May 3, 2012

gollywop wrote:

Iliah Borg wrote:

Another reason is that muted ETTR colors and low JPEG contrast impair focusing speed

Try b/w

Excellent idea.

It's the slick way to go. Lose those green-tint blues forever ! I've been saying it for a while now. Maybe now that the eminent Iliah Borg has revealed using B&W at times himself, it will catch on. It kills the (playback only) "blinkies" on the GH2 - but is reported not to do that on an Olympus model for the preview "blinkies" (I forget which one at the moment - it was not an E-M5). Try it !!

Reply   Reply with quote   Complain
Detail Man
Forum ProPosts: 14,837
Like?
Re: Configure UniWB for Dummies
In reply to Bob Tullis, May 3, 2012

Bob Tullis wrote:

Is there such a writeup?

In the case of kenw 's DMC-G1, all one has to do is aim the cam at a very bright lamp that will "blow the smithereens" out of all of the RAW-channels - and caibrate Custom WB to that. The result is perfect R=1.0, G=1.0, B=1.0.

That technique does not work on the GH2 (which senses the high light-levels, and refuses to calibrate the Custom WB as a result).

Try that "brute-force hammer" technique on your E-M5's, guys ... (maybe) it works like the G1 ?

Reply   Reply with quote   Complain
gollywop
Veteran MemberPosts: 5,304
Like?
Re: E-M5 ACR Baseline Exposure, UniWB, and ETTR
In reply to micksh6, May 3, 2012

micksh6 wrote:

gollywop wrote:

Indeed, I am still not convinced the jpeg engine plays any part in the cdaf at all. I would appreciate anyone who knows how it works to chime in.

Nobody knows exactly how it works so it's unlikely that someone will tell you. You won't find any official documents. Manufacturers will never disclose how their cameras work in details. Reverse engineering is sometimes the way to find out.

I fear you are correct here. But there may be some general principles involved that we can know. For example, it appears that the sensor image data are employed for cdaf. But just how?

Since cdaf speed has always been an issue, it would seem likely that the process has evolved to make use of those data as efficiently and quickly as possible, and that, it would seem to me, would preclude encumbering the operation with unnecessary calculations like WB multipliers, or, for that matter, with other camera tonal settings -- even if these items are added in a less critical manner for display purposes. The two processes needn't be linked. Likely these things will vary from manufacturer to manufacturer, but you would think some of these notions would emerge even if the very specifics did not.

As a matter of interest, I have just done some tests with my E-M5 with s-af, and there is absolutely no difference in the focusing action or speed with AutoWB, UniWB, muted, vivid, high/low contrast. I haven't messed around with c-af.

You normally won't notice S-AF autofocus difference unless you focus in really low light. With decent light AF speed gain is measured in tens of milliseconds, it's beyond human abilities no notice it. C-AF will show the difference. S-AF in normal light will show too, when you shoot fast moving subjects, but only statistically, when you look at percentage of sharp photos.

I pretty much explained how it most likely works from point of view of image processing software engineer. You may believe it or not, I don't have to convince you. I am all ears if someone provides better explanation.

It's not a matter of belief; it's a matter of speculation for all of us. You may well be correct, as I've indicated above, but I do question whether all the operational bells and whistles needed for the image processing, but are not needed for cdaf, are employed in the focusing algorithm.

I too would be delighted if someone who really knows what's what could chime in here or in your new thread to help set the matter straight from a theoretical perspective. Shy of that we need good, reproducible experimental data from which meaningful inferences can be made. What we don't need is further speculation. I think most of the speculation of significance is already on the table from you and DonParrot.

I certainly do not claim to know for sure, but I wouldn't be the slightest bit surprised if there were dedicated circuitry to assess the cdaf, quite apart from the jpeg engine and having nothing directly to do with the full image. There is nothing about the existence of LV that precludes this possibility.

Yes, there is. It takes at least 4096x3 array of coefficients to load into DSP in order to change picture settings from live view to CDAF optimized. Doing that may easily break AF speed requirement of 120 fps rate on some hardware.

Ok; let's see if we can push this further to get some salient facts.

-- hide signature --

gollywop

-----------

Reply   Reply with quote   Complain
gollywop
Veteran MemberPosts: 5,304
Like?
Re: Configure UniWB for Dummies
In reply to Bob Tullis, May 3, 2012

Bob Tullis wrote:

Is there such a writeup?

Yes, Bob. It's at

http://forums.dpreview.com/forums/read.asp?forum=1041&message=41245342
--
gollywop

-- hide signature --

Reply   Reply with quote   Complain
cosmoxl
Contributing MemberPosts: 581Gear list
Like?
tried it
In reply to Detail Man, May 3, 2012

tried the WB on a bright light with the E-M5. the camera denied it.

 cosmoxl's gear list:cosmoxl's gear list
Panasonic Lumix DMC-FZ50 Olympus OM-D E-M5 Olympus E-M1 Olympus M.Zuiko Digital ED 9-18mm 1:4.0-5.6 Olympus M.Zuiko Digital ED 75-300mm 1:4.8-6.7 +5 more
Reply   Reply with quote   Complain
cosmoxl
Contributing MemberPosts: 581Gear list
Like?
try this
In reply to dcassat, May 3, 2012

the color that gave me UniWB was RGB 255, 55, 211 with my E-M5. you might start there.

 cosmoxl's gear list:cosmoxl's gear list
Panasonic Lumix DMC-FZ50 Olympus OM-D E-M5 Olympus E-M1 Olympus M.Zuiko Digital ED 9-18mm 1:4.0-5.6 Olympus M.Zuiko Digital ED 75-300mm 1:4.8-6.7 +5 more
Reply   Reply with quote   Complain
gollywop
Veteran MemberPosts: 5,304
Like?
Re: E-M5 ACR Baseline Exposure, UniWB, and ETTR
In reply to micksh6, May 3, 2012

micksh6 wrote:

gollywop wrote:

micksh6 wrote:

gollywop wrote:

The WB doesn't come into play until after the shot is taken and the raw data are processed in-camera by the jpeg engine.

Just noticed this phrase. This is the error in logic. We just recently discussed green live view with UniWB, right? Live view works before the shot is taken, it uses JPEG engine and WB is a part of it.

There is no question but what the WB is applied to the LV image. But that does not mean that it is part of the cdaf algorithm, which, as I understand it, reads from a number spots directly off the sensor. So it is not shown there is any error in logic. The error could be (and I suspect is) in the synthesis "and WB is a part of it."

Indeed, I am still not convinced the jpeg engine plays any part in the cdaf at all. I would appreciate anyone who knows how it works to chime in. I certainly do not claim to know for sure, but I wouldn't be the slightest bit surprised if there were dedicated circuitry to assess the cdaf, quite apart from the jpeg engine and having nothing directly to do with the full image. There is nothing about the existence of LV that precludes this possibility.

And you were absolutely correct in your earlier post when you suggested the likelihood that I would not find the citations you gave to provide significant evidence.

As a matter of interest, I have just done some tests with my E-M5 with s-af, and there is absolutely no difference in the focusing action or speed with AutoWB, UniWB, muted, vivid, high/low contrast. I haven't messed around with c-af. I'll do that when I get some time.

CDAF is is certainly different topic from UniWB so I created another thread
http://forums.dpreview.com/forums/read.asp?forum=1041&message=41406241

Good idea; I'll follow it with interest.

And, BTW, thank you for your post regarding UniWB on E-M5. I will certainly use UniWB when I get this camera.

You're welcome; and it's not all that difficult to set up with the E-M5. You're really going to enjoy that camera when you get it.

PS. Measuring EV difference between UniWB and AutoWB is less work than shooting grey card to determine WB. And it also doesn't require carrying grey card.

Well, I don't know about this. Shooting a gray card becomes second nature and quick. And, once done, you get the appropriate WB information, which does not always come with AutoWB. With all my cameras -- including the E-M5 -- I've checked out the AutoWB by shooting scenes using it and also establishing the WB for that lighting via a gray card. Sometimes the AutoWB is really excellent, but in scenes (such has closeups of flowers) that contain strongly saturated colors with no major neutral regions, the AutoWB can be horrifically misleading. I've found no camera's AutoWB able to avoid this; indeed it doesn't seem possible to do so on the face of it.

Of course, AuotWB is also inappropriate for shooting sunsets, but that's a different issue. In those shots, no gray card or AutoWhatever is going to help. That's a time to use an appropriate preset or estimate an appropriate WB in PP.

-- hide signature --

gollywop

-----------

Reply   Reply with quote   Complain
Bob Tullis
Forum ProPosts: 25,280Gear list
Like?
Re: try this
In reply to cosmoxl, May 3, 2012

cosmoxl wrote:

the color that gave me UniWB was RGB 255, 55, 211 with my E-M5. you might start there.

Thank you. I just haven't made the time to focus and do it - this is a helpful start with Gollywop's link.

-- hide signature --

...Bob, NYC
http://www.bobtullis.com

"Well, sometimes the magic works. . . Sometimes, it doesn't." - Little Big Man
.

 Bob Tullis's gear list:Bob Tullis's gear list
Olympus E-M1 Rokinon 7.5mm 1:3.5 UMC Fisheye CS Voigtlander Nokton 42.5mm F0.95 Voigtlander Nokton 17.5mm F0.95 Aspherical Olympus 12-40mm F2.8 +17 more
Reply   Reply with quote   Complain
Keyboard shortcuts:
FForum MMy threads