Forgot Password
Pentax Camera Forums Home
 

Reply
Show Printable Version Search this Thread
05-20-2008, 04:56 PM   #61
Pentaxian
falconeye's Avatar

Join Date: Jan 2008
Location: Munich, Alps, Germany
Photos: Gallery
Posts: 6,862
QuoteOriginally posted by PentaxPoke Quote
All exposure values are exactly the same in these two shots. Both are RAW conversions. Note that the wagon is supposed to be RED!
I so much really would like to agree. Helas...

I only now have discovered the thread and would like to contribute 3x2 cents...


(1) This example (the wagons)

I compared channel histograms on both samples. The red channel on both samples is heavily clipped. By about the same amount. It is by pure accident that one came out slightly better than the other. Unfortunately statistically insignificant for making statements about EDR.

The difference in histograms in the cameralabs sample was statistically significant, though and I guess it was shot in JPEG (cf. below).


(2) EDR in Raw mode

It seems to be an accepted fact by now that EDR has no effect in RAW mode. Chasseur d'Images claimed that it enables a mysterious 14 Bit mode but this has been proven to be wrong. Gordon's original work about EDR and RAW is here:

Re: K20D: EDR *does* affect RAW images.: Pentax SLR Talk Forum: Digital Photography Review


(3) EDR in JPEG mode

This is my standard theory of what EDR in JPEG does ... and it is surviving since 2 months already and is backed up by all sources I am aware of, incl. Gordon's work cited above

EDR puts the camera internally at one ISO step below what it displays and at what it actually assumes in its exposure settings.

So, when at ISO200 in EDR, the camera exposes at ISO200 but captures at ISO100. BTW, this explains why ISO100 becomes unavailable as a mode.

This should give an underexposed image, right? But now, EDR kicks in: It will boost the shadows up to the mids by one full stop (hence the possible extra shadow noise -- note however that the shot was taken at ISO100 so it won't be as bad altogether). And it won't boost the highlights (gaining one EV in captured dynamic range) with bright colors boosted somewhere in between. This is called tone mapping in HDR processing. That's it. No mystery. What you do in RAW anyway, assuming you decided to take an underexposed ISO100 shot of a strong contrast subject.

- The drawback: Flatter contrast and more shadow noise than possible at ISO100 w/o EDR.
- The advantage: Lower risk of blown highlights or ruined specular colors.
- Personally, I use EDR with JPG and at daylight (indoor and outdoor) only.

If anybody can correct me and show that EDR is a more magic thing then I will be glad to learn about!


Last edited by falconeye; 05-20-2008 at 05:01 PM.
05-20-2008, 05:43 PM   #62
Veteran Member
WMBP's Avatar

Join Date: Dec 2006
Location: Dallas, Texas
Photos: Gallery
Posts: 2,496
QuoteOriginally posted by falconeye Quote
I so much really would like to agree. Helas......
And the ball bounces back to the other side of the court once again!

Falconeye, I read that thread at dpreview.com when it first appeared but didn't really understand it very well. I think I understand it better now, but it does seem that all of this is reverse engineering - clever folks trying to make sense of what they see in the output, when the Pentax engineers could save us all a lot of trouble by just telling us what's going on.

So now it seems that EDR does nothing in raw mode? Does the user's manual for the K20D say that? I have read the manual, once, as attentively as I could, given the wretchedness of the thing, and I don't recall getting that. Indeed, if this were true, and if I had programmed the firmware, then the EDR option simply wouldn't be available in raw mode. Raw+, yes, but plain raw (which is almost all I shoot), no.

Now this raises a question I've never asked myself before. When you view the histogram on the camera right after taking a shot, is that the histogram of the jpeg preview, or is it the histogram of the raw capture? The reason I ask is, that I took some EDR test shots myself a week ago, and I could see - or thought I saw - a small but detectable difference in the highlights with EDR on and EDR off.

Sigh.

Will
05-20-2008, 06:33 PM   #63
Veteran Member
PentaxPoke's Avatar

Join Date: Jan 2008
Location: Oklahoma
Posts: 1,411
Original Poster
I don't understand how you say it does nothing in RAW mode and then post a thread that "proves" that it does.

Falconeye, do you own a k20d? Have you actually tried EDR? I shoot in RAW only mode and EDR has an effect. I convert to jpeg in ACR, so how can you say that EDR has no effect in RAW, only jpeg? How can something only affect jpeg when I don't even shoot in jpeg!?!

Regarding item 1, I disagree that it is "luck." There was more color information available in the EDR ON image. Even though both images were clipped (like I planned) there is still more information in the EDR ON image. That is the whole point!!! It behaves very similar to what Ken Rockwell described in the D300 thread. I could prove it with the histograms of the original RAW images (that you don't have) but there is no need; look at the images with your eyes! There is clearly more light and color information there, and specular highlights are not as blown out with EDR ON. Just take a look at the highlights on the blue wheels for example.

Last edited by PentaxPoke; 05-20-2008 at 07:41 PM.
05-20-2008, 07:15 PM   #64
Junior Member




Join Date: Apr 2008
Location: Prague
Posts: 42
Most questions from this thread, including this one
QuoteOriginally posted by WMBP Quote
When you view the histogram on the camera right after taking a shot, is that the histogram of the jpeg preview, or is it the histogram of the raw capture?
can be quite easily answered by simple practical tests. But somehow I feel many people prefer endless talks instead, so I will not spoil the fun by answering it

05-21-2008, 03:54 AM   #65
Pentaxian
falconeye's Avatar

Join Date: Jan 2008
Location: Munich, Alps, Germany
Photos: Gallery
Posts: 6,862
Well, seem's I've upset a number of people here. Not intended. I only wanted to be helpful. And yes, I do own a K20D myself and what I said is backed from my experiments I did with it, too. Anyway, let's go thru some of the back fire I received

QuoteOriginally posted by PentaxPoke Quote
I don't understand how you say it does nothing in RAW mode and then post a thread that "proves" that it does.
The thread proves what I said (except that I was somewhat brief, cf. below). The thread's title is misleqading, though, because Gordon's work was in reply to an original post which he proved wrong.

QuoteOriginally posted by PentaxPoke Quote
I shoot in RAW only mode and EDR has an effect. I convert to jpeg in ACR, so how can you say that EDR has no effect in RAW, only jpeg? How can something only affect jpeg when I don't even shoot in jpeg!?!
Well, maybe, I was too brief here. Before tone mapping, EDR does the same in RAW and JPEG modes, namely: "EDR puts the camera internally at one ISO step below what it displays and at what it actually assumes in its exposure settings."

However, independent of EDR, this can always be done in RAW mode by setting the ISO to 1/2 of what is intended and dial in EV compensation of -1EV (which doubles the well capacity and increases the read-out noise). I should have been clearer in what I said. Sorry about this.

What I should have said is that EDR in RAW has no effect other than available thru the dials.

QuoteOriginally posted by PentaxPoke Quote
There was more color information available in the EDR ON image. Even though both images were clipped (like I planned) there is still more information in the EDR ON image.
I can see the difference as everybody else. The second image looks better. It may be that the subtle difference is due to EDR, maybe not. I was only saying that the subtle difference in histograms of pictures as posted is, unfortunately, not statistically significant. I attach the histograms of both. In JPEGs from EDR, the difference is much more pronounced.

Attachments: EDR off/on. The highlights may be stronger in #1, but clipped at the same point (EDR on at only 1 RGB value later -- which I call statistically insignificant even if it has a clear visible effect).

@WMBP: You are right with your comment. The user's guide only states what is obvious after using the camera for a couple of days. An effect of outsourcing documentation rather than let the engineers write a draft and only outsource the polishing of it. Only very few companies do it differently now.

As for EDR in RAW only. I think EDR wasn't intended to be used in conjunction with RAW. But of course, it does what it does, no need to disable it. Esp. as you wouldn't be able to mimic it in Auto-ISO mode.


And sorry again for intervening into this discussion.

Last edited by falconeye; 06-15-2011 at 05:27 AM.
05-21-2008, 08:48 AM   #66
Veteran Member
WMBP's Avatar

Join Date: Dec 2006
Location: Dallas, Texas
Photos: Gallery
Posts: 2,496
QuoteOriginally posted by falconeye Quote
And sorry again for intervening into this discussion.
Falconeye, please don't apologize. I found your contribution very interesting and valuable. I'm just an interested spectator in this thread. All I want is to learn what I can learn about this feature, in particular, I'd like to learn if it actually does anything that I can't do on my computer in post-processing. And I would note that PentaxPoke also seems interested only in figuring out the truth here. He seems to have changed his mind about EDR in the course of this thread and I suppose he'd be willing to change it again. But in the absence of authoritative info, we're all left to make our cases one way and the other and see what happens.

I'm still confused, but I am trying my best to follow the back and forth here and enjoying the discussion very much.

Will
05-29-2008, 08:21 PM   #67
Veteran Member




Join Date: Jun 2007
Location: Owego, NY
Posts: 976
QuoteOriginally posted by WMBP Quote
So now it seems that EDR does nothing in raw mode? Does the user's manual for the K20D say that? I have read the manual, once, as attentively as I could, given the wretchedness of the thing, and I don't recall getting that. Indeed, if this were true, and if I had programmed the firmware, then the EDR option simply wouldn't be available in raw mode. Raw+, yes, but plain raw (which is almost all I shoot), no.
EDR does something in raw mode - it underexposes by one stop, nothing else.

After noticing that most of my pictures (taken with EDR on) from the past weekend were severely underexposed, I did the following:

Put the K20D in manual mode
Set ISO to 400, D-Range 200% to ON
Aim it at a white sheet of paper
Set aperture and shutter so that camera reads a bit over 1 EV overexposure (since the sheet is pure white and not grey)
Take a shot, make sure it's not clipping (histogram spike about 1/8 from max, no clipping)
Turn off EDR
Take another shot, make sure it is still not clipping
Change ISO to 200, leave EDR off
Take one last shot
Load the three PEFs into UFRaw with exposure compensation set to Auto. Note the exposure compensation chosen by UFRaw for each PEF.

Theory: ISO 400 and 200 with EDR off should differ by one stop since shutter, aperture, subject, and lighting were all identical. ISO 400 with EDR on will be the same as ISO 200 with EDR offif the theory that underexposing the PEF by 1 stop is true.

Results:
ISO 400 EDR ON: UFRaw chose +1.24 stops exposure compensation
ISO 400 EDR OFF: UFRaw chose +0.24 stops exposure compensation
ISO 200 EDR OFF: UFRaw chose +1.24 stops exposure compensation

So if you are shooting RAW only, EDR does only one thing - underexpose your shot by one stop.

Any suggestions how I can provide my complete dataset (the three PEFs) conveniently if anyone wants to see them for confirmation? I'd post them right here except they are around 15M each (to be expected)

Edit: The manual should either clearly document this or the function should be locked out in RAW-only mode, since it has zero benefit.
05-29-2008, 08:40 PM   #68
Veteran Member
PentaxPoke's Avatar

Join Date: Jan 2008
Location: Oklahoma
Posts: 1,411
Original Poster
Entropy,

I'm not so sure that it is as simple as that. Wouldn't that mean that the ISO 400 EDR on and ISO 400 EDR off histograms would be shifted by 1 stop (before any PP)? I don't see that at all in my shots, or any other tests such as the Cameralabs test. Maybe that would be the case for a uniform all-white sheet? Is auto post processing the best way to evaluate what the camera is doing?

05-29-2008, 10:43 PM   #69
Veteran Member




Join Date: Jun 2007
Location: Owego, NY
Posts: 976
QuoteOriginally posted by PentaxPoke Quote
Entropy,

I'm not so sure that it is as simple as that. Wouldn't that mean that the ISO 400 EDR on and ISO 400 EDR off histograms would be shifted by 1 stop (before any PP)? I don't see that at all in my shots, or any other tests such as the Cameralabs test. Maybe that would be the case for a uniform all-white sheet? Is auto post processing the best way to evaluate what the camera is doing?
Histograms displayed in the camera are NOT for the raw data, but for the full resolution JPEG (if produced for RAW+) or for the reduced resolution preview JPEG embedded in the RAW image.

In the case of EDR, the histogram shown is likely one that has had the EDR postprocessing applied (to generate the preview image). I had been wondering for a while why digikam's previews (based on the embedded JPEGs) were not underexposed by a stop when UFRaw's previews (calculated from the RAW data using a fast half-resolution interpolator) were.

As to the use of an automatic mode in UFRaw, in this case, the goal was to see what sort of correction the automatic mode used to achieve what it thought was correct exposure, given three files with identical shooting conditions except for ISO settings (Note that the camera was in manual mode for these). The end results are in my previous post, with an end confirmation that the resulting post-conversion histograms had their spike very close to each other.

Edit: What RAW converter are you using that gives a pre-conversion histogram? Every one I have seen only displays the estimated post-conversion histogram for the purposes of assisting with exposure compensation adjustment during conversion.
05-29-2008, 11:15 PM   #70
Veteran Member
PentaxPoke's Avatar

Join Date: Jan 2008
Location: Oklahoma
Posts: 1,411
Original Poster
I am using ACR. Also, I am not aware of any PP of EDR. I don't even know if any RAW converters are even "aware" that EDR was used. I am guessing the effect of EDR occurs in the camera, and there is no difference between the processing of the files.
05-30-2008, 05:14 AM   #71
Veteran Member




Join Date: Jun 2007
Location: Owego, NY
Posts: 976
QuoteOriginally posted by PentaxPoke Quote
I am using ACR. Also, I am not aware of any PP of EDR. I don't even know if any RAW converters are even "aware" that EDR was used. I am guessing the effect of EDR occurs in the camera, and there is no difference between the processing of the files.
Nope, the RAW converters are not. As a result, if you take two shots of the same scene, one with EDR on and one with EDR off in the exact same lighting with the same shutter and aperture, the shot with EDR on should come out one stop underexposed if no correction is applied. If you apply correction to get the histogram of the EDR On picture to match that of the EDR Off picture, you will find that the EDR On picture needed an extra stop of brightness added. (I don't recall if ACR does brightness correction in +/- EV, UFRaw does so it's easy to see a difference of 1 stop.) Also, the EDR On picture at ISO N will have the same histogram as an EDR Off picture at ISO N/2 if you apply the same brightness correction to each.
05-30-2008, 06:10 AM   #72
Pentaxian
falconeye's Avatar

Join Date: Jan 2008
Location: Munich, Alps, Germany
Photos: Gallery
Posts: 6,862
QuoteOriginally posted by Entropy Quote
ISO 400 EDR ON: UFRaw chose +1.24 stops exposure compensation
ISO 400 EDR OFF: UFRaw chose +0.24 stops exposure compensation
ISO 200 EDR OFF: UFRaw chose +1.24 stops exposure compensation

So if you are shooting RAW only, EDR does only one thing - underexpose your shot by one stop.
Hi Entropy,

your findings are in line with that of others, most notably Gordan's work I was referring to above.

However, EDR does do another thing as well ... it goes down one ISO step internally.

(Which is why I said: half the ISO and use -1EV compensation to mimic it.)

Since you didn't clip, you couldn't see it (going down one ISO step doubles the well capacities). Since you didn't shoot shadows, you coudn't measure the slightly added shadow noise due to the higher read-out noise at the lower ISO.

According to some sources, ACR and LightRoom are aware of the implicit underexposure to expect from EDR raw shots and correct for it when seeing the EDR meta flag.


BTW.
EDR mode with RAW isn't a setting to be avoided as such (as you suggested elsewhere). Because ISO N/2 & -1EV cancels out each other you still get about the same image quality (the lower raw values are due to the read-out values divided by 2 from the lower ISO -- the number of collected electrons is the same, though). You still trade higher read-out noise vs. higher well capacity, or in photographic speech, somewhat higher shadow noise vs. more headroom for unclipped highlights.

In bright sunshine, using ISO100 and -1EV compensation may be a common setting. ISO200/D-Range is just a short-cut for it

Last edited by falconeye; 05-30-2008 at 06:27 AM.
05-30-2008, 06:15 AM   #73
Veteran Member
lbam's Avatar

Join Date: May 2008
Location: Toronto
Photos: Gallery
Posts: 443
Shooting in RAW with EDR on, and viewing the results in Lightroom, I see a distinct flat spot on the right hand side of the histogram...
Just my thoughts...
05-30-2008, 11:36 AM   #74
Inactive Account




Join Date: Jul 2007
Location: London
Posts: 1,067
Hi lbam

That 'flat spot' resembles Obi-Wan Kenobi in the Gobi desert, by the look of it.......lol !

Best regards
Richard
05-30-2008, 05:44 PM   #75
Veteran Member
lbam's Avatar

Join Date: May 2008
Location: Toronto
Photos: Gallery
Posts: 443
Errr.. you lost me there
Reply

Bookmarks
  • Submit Thread to Facebook Facebook
  • Submit Thread to Twitter Twitter
  • Submit Thread to Digg Digg
Tags - Make this thread easier to find by adding keywords to it!
camera, detail, dslr, experiences, feature, k20d, opinions, photography, pictures, range, shadow
Thread Tools Search this Thread
Search this Thread:

Advanced Search


Similar Threads
Thread Thread Starter Forum Replies Last Post
expanded dynamic range in the K20 insulinguy Pentax DSLR Discussion 0 09-03-2009 01:25 PM
K2000/K-M Expanded Dynamic Range lawsonstone Pentax DSLR Discussion 7 07-03-2009 07:10 PM
dynamic range K20D dafiryde Pentax DSLR Discussion 4 01-08-2009 10:51 AM
K20d Dynamic Range? Don Boyd Pentax DSLR Discussion 2 04-07-2008 04:07 AM
K20D expanded dynamic rage test request Oceanic Pentax DSLR Discussion 2 03-29-2008 12:41 PM



All times are GMT -7. The time now is 07:26 PM. | See also: NikonForums.com, part of our network of photo forums!
  • Red (Default)
  • Green
  • Gray
  • Dark
  • Dark Yellow
  • Dark Blue
  • Old Red
  • Old Green
  • Old Gray
  • Dial-Up Style
Hello! It's great to see you back on the forum! Have you considered joining the community?
register
Creating a FREE ACCOUNT takes under a minute, removes ads, and lets you post! [Dismiss]
Top