[PentaxPoke]

I would like to get the opinions of other k20d users on the 200% expanded dynamic range setting, based on experiences shooting in that mode. I know there was a thread somewhat similar to this in "rumors" but I think this is the best venue. I have run many tests myself and I just don't see the effect yet. I have read two different things:

- Some say it reduces highlight clipping
-Others say it increases shadow detail

I'm not sure how it does both, and I just haven't seen it myself in any pictures I have taken no matter how "contrasty" the lighting is. I am a big believer that one should post pictures to make a point, but trust me on this, there is no reason for me to waste bandwidth on this site posting pictures, because I see no discernable differerence on and off in my pictures.

Does anyone else have some samples on and off of the same image that really show the value of this feature? I have seen the Pentax video but it really doesn't show it (other than to point out a potential of increased shadow noise). I have searched on the internet, and the closest I have found is this test:
http://www.markcargill.co.uk/media/K20D%20review/Pentax%20K20D%20review.pdf

on pages 2-3. The example makes no sense to me because I see LESS shadow detail rather than more. Is it possible this "feature" is a marketing gimick? If it truly increased dynamic range, why wouldn't it be on all the time, instead of being an option?

Please note: I am not asking what it does, or what it is supposed to do. That information is widely available. I am asking you for your experiences with this feature, and your opinion on how it made your photo better.

---

Last edited by PentaxPoke; 05-02-2008 at 05:50 PM.

[Sergei]

I've only had my K20D for 2 weeks, and have tested the feature to see the effectiveness.
I have discerned about 1.5 stops IMO, with definite clipping reduction in highlights, and some gain in shadow detail. Depends very much on subject matter.
Shooting a wedding tomorrow, so will give it a good workout under different conditions.

[PentaxPoke]

Interesting. Any chance of posting examples showing the 1.5 stops improvement and better shadow detail? Maybe I am doing something wrong, or it is not working.

[Sergei]

Interesting. Any chance of posting examples showing the 1.5 stops improvement and better shadow detail? Maybe I am doing something wrong, or it is not working.

Certainly,
Will try to dig out some previous tests tonight, but will carry out some specific tests tomorrow at a wedding shoot.

[DAZ]

You might see something here. https://www.pentaxforums.com/forums/pentax-dslr-discussion/22265-what-differe...ally-make.html

[PentaxPoke]

DAZ,

Yes I had seen that post. I should have refered to that one in addition to the one in the "rumors" forum. You have a nice example that if I can paraphrase: shows no improvement in shadow detail, and there is a slight improvement in the highlight clipping? Your attempt is better than most I have seen to show the advantage of the EDR.

I am just not seeing the 200% DR improvement. Especially at the shadow end. So is the bottom line this feature just works at the extreme end of exposure to limit clipping? I wonder why Pentax warns of increase in shadow region noise if it doesn't do anything there. Why is it not permanently on if it is nothing but a clipping safegard? Heck you showed that you have to go to 100% crops to even see a slight effect.

I am so confused

[WMBP]

Please note: I am not asking what it does, or what it is supposed to do. That information is widely available. I am asking you for your experiences with this feature, and your opinion on how it made your photo better.

Actually, I'm still unclear about what it is supposed to do.

I'm also not sure that it actually does anything. I've tried to do some test shots, but I think I need to set the tests up inside with artificial light. When I try to shoot outside in bright daylight, clouds keep rolling around and changing the light and I'm not sure whether it's the EDR or the clouds that's making the difference.

I read (and contributed briefly to) the thread that DAZ links to. I can see what's being claimed there. But the feature's effects remain fairly iffy to me.

Will

[PentaxPoke]

Actually, I'm still unclear about what it is supposed to do.

I'm also not sure that it actually does anything. I've tried to do some test shots, but I think I need to set the tests up inside with artificial light. When I try to shoot outside in bright daylight, clouds keep rolling around and changing the light and I'm not sure whether it's the EDR or the clouds that's making the difference.

I read (and contributed briefly to) the thread that DAZ links to. I can see what's being claimed there. But the feature's effects remain fairly iffy to me.

Will

At least I am beginning to feel better that my k20d is not broken. Now if only I could figure out why that "feature" is there in the first place! I look forward to seeing Sergei's test shots at a wedding. That is a "critical" dynamic range environment where you would think EDR could pay off: White dress, black tux, backlighting through stained glass, dire consequences if the shot is blown, etc. The trick will be to get the same shot with EDR on and off to show us the difference. No pressure Sergei!

[Sergei]

O.K. I tried to troll my folders for the countless trashy test shots that one does as soon as they get a new camera, and I have to admit that I could not match a pair that clearly demosttrates the effect of EDR.
The problem is that in a practical situation, you are not likely to be shooting mannual exposure, so all my initial tests were in SV mode.
The EXIF data unfortunately does not help in stamping the respective states.
So I did another quick assesment this afternoon, mannual exposure and focus.
WB locked to Daylight, 200ISO.
Admitedly here the lighting included a lot of bounce fill, so the contrast ratio is not that extreme, even though the subject is backlit.
So my conclusion from this particular exercise is that,
most of the gain is in the highlight area, clearly visible in the curtain area to the right of the caref.
The shadows in this case are difficult to appreciate, however there is a lot of bounce fill as already noted.
So tomorrow afternoon, the wedding shoot will present a real case scenario, I will shoot in SV mode, as is my usual practice.

No presssure

O.K. so now for the attachments

[Sergei]

Oh ... and the Left is OFF
and Right is ON,

very subtle midrange detail difference on the tiles is just noticeable,
no USM or PP of any kind, just crops and reduced in size.

[Peacekeeper]

I can also see more detail in the white curtain area that is blown out on the left and the wall txture has more contrast to.

PK

[Confused]

Hi Sergei

Thanks so much for posting these two interesting photos, because I've been waiting some while to see a good example clearly illustrating the visible benefits of EDR. The shadow in the floor-tiles exhibits much greater subtlety and seems far more pleasing in the second photo and I can clearly see that the second image has retained far more detail, particularly in the fine mesh of the curtain material, where the highlights have obviously blown out in the first image. Although employing a genuine HDR technique in these circumstances might result in still greater depth being resolved, the EDR mode definitely appears to work satisfactorily in this instance.

Many thanks
Richard

[PentaxPoke]

Great test Sergei. Thank you for taking the time to do this. The differences are very subtle. I almost see no difference. I don't know if I attribute the difference in the wall texture to EDR, or subtlety of focus.

So folks, does anyone see a 200% change in DR? What's the consensus? Leave it on all the time? I don't see any increase in shadow noise.

[DAZ]

There are severely issues with looking at dynamic range of a camera. The first is you have to have a subject that has enough dynamic range to show something. This means that one of the photos has to have one or both ends in the black or clipping and not in the other photo. If it is not then you have not exceeded the dynamic range of that mode and you will not have anything to see. A lot of scenes just don’t have enough dynamic range to see anything.

Most camera systems clip at about the same point. Also clipping is easy to determine so it take less of a judgment so it is used as the referents point for one end of the dynamic range. This means that everyone talks and argues about the dark end. The dark end is harder to determent as you have to separate were the noise floor is from the last little bit of detail is. There is just a lot less information in the dark end to find.

Because the dark end is limited by noise there are some very big problems (like quantum noise) with trying to extend to dark end of the dynamic range. Because of the liner way the sensors work there are less bits (interesting things) to get from this end. One could make a camera samples somehow more at the dark end but noise tends to get in the way. At the high end if this is extended there is more information and a lot less noise so this is probably the end that is going to give more if you extend the dynamic range.

In the days of film we would expose for the darker parts because film doesn’t clip like digital. In digital we have to expose for the light end (expose right) and pull up any dark parts we can/want. The farther you can go right the more, in theory, you can get the dark parts.

To see detail at something right at the end of the high end you need something that has detail. White window probably not but white lace on a white wedding dress yes. Some things like white on white is just hard to see. If you just totally clip the white dress you will not see any lace but to really see the lace you have to use something like curves to pull out the detail and make the lace stand out. This works best with RAW as the additional detail may get lost in JPG.

Something like this Flickr Photo Download: CanadianGoosePortraitDAZ_0504 can possible be of use with 200% dynamic range but a lot of day to day photos may not.

When I was testing I used manual to eliminate any influents from the metering system and have control over when I clipped. You can just use any control mode you wish and just use the exposure adjust like you normally do. To see any really dynamic range change you must be clipping with it off or there is nothing to see.

So I think there is more range but you have to have something with the range to see it and expose and PP to get it. How much more I am not totally sure. I think it is ½-1 in RAW and 1-1.5 over JPG.

DAZ

[PentaxPoke]

OK I ran my own tests today. I think my results are consistent with DAZ. Here is the setup:

Bright late afternoon sun
Shoot on manual
ISO 200 for each shot. Exposure settings identical
Shoot DR on and DR off
All images shot in PEF RAW

The shots are of a neighbors 4 wheeler. Lots of contrast. Note the garage door in the background showing lots of white, and the open door next to it with dark.

When I made the shots with EDR on and EDR off, the only thing I noticed after the shot was that the garage door "blinked" in the LCD due to clipping when EDR was off. When it was on, I didn't get the clipping.

The difference in the shots was not apparent, so in ACR, I cranked up the recovery slider to 100% which brought the highlights down. I also brought up the fill light to 50% to see about bringing out shadow detail.

Summary of results: The main difference can be seen if you zoom in to the garage door detail. Most of it is preserved in the EDR on case, where it is mostly clipped with EDR off. No difference in shadow detail or noise could be seen. Here are the photos:
k20d EDR test - a set on Flickr

I'm still at a loss to see the 200%. Wouldn't a 200% increase mean a 2-stop increase in DR? There is not even a 1-stop increase as far as I can tell, on either the high or the low end. If EDR were going to show itself, it would have done so here. The effect is very subtle, bordering on unobserveable.
Comments?