exposure compensation

Report bugs and get support here

Moderator: Stas Yatsenko

firexball
Posts: 21
Joined: 07.07.2007 18:10

exposure compensation

Unread postby firexball » 24.08.2007 17:11

Hello,

today I recognized, that the exposure compensation of Helicon Filter doesn't do the same that other software do!?! In my opinion, HF exposure compensation has too much contrast.
If i do an exposure compensation of +1.5 in DPP, the results are close to HF with +50% exposure compensation and -50% contrast. Can anyone confirm this?

Is there a global definition, how exposure compensation have to be calculated? If yes, does HF really use this? (Maybe the results of DPP are wrong?)

Next step I want to do, is to make the same photo with different exposures (-4, -2, 0, +2, +4) and compare them with the HF results.

Greetings firexball

firexball
Posts: 21
Joined: 07.07.2007 18:10

Unread postby firexball » 27.08.2007 00:57

OK, now I did some more testing and prepared the following pictures to explain what I mean:

These are the basic shots I've made. On the left side there are the underexposured Pictures with -2 and -1. In the middle there is a Picture with the right expoure and on the left there are the +1 and +2 exposures.

Image

Ok, now I've tried to correct all Pictures to the correct (middle) exposure. First I did this by using exposure silder in DPP.

Image

All Pictures are looking ok, except the +2 Picture.

In the next step I tried the same with the exposure slider in Helicon Filter.
Because there are no aperture steps on the slider, I've tried as good as possible.

Image

I think you can see the difference. Only the -1 Picture is acceptable. But as you can see there is a bit too much contrast in that image. In the -2 image this effect is even stronger. (the Background is ok, but the dog is too dark!)
The +1 and the +2 images are not correctable with this slider in HF.

In my opinion the exposure slider should correct the exposure of an image as realistic as possible. In the ideal case there shouldn't be a difference.
Also, the exposure silder should show the number of aperture steps and not an virtual value.

I hope my little analysis could help you.

Greetings firexball

User avatar
Dan Kozub
Posts: 1157
Joined: 24.03.2004 18:14

Unread postby Dan Kozub » 28.08.2007 21:34

Hello firexball!

Thanks for the interesting test. Exposure slider in Helicon Filter produces linear effect on all pixels.

When using DDP do you apply exposure compensation to RAW file or to the JPEG or Tiff file? This can make the difference.

firexball
Posts: 21
Joined: 07.07.2007 18:10

Unread postby firexball » 29.08.2007 22:28

Hello!

Yes, I really applied exposure compensation to the Raw-File (which is linear).

After some research about linear workflow during the last days, I now understand, why HF can't do proper linear exposure compensation after "normal" Raw-conversion.

So I tried the test above with DCRAW and the 16bit linear option and got much better results with HF. The only problem is, that this is an linear image and not suitable for printing,...!

Maybe there could be the possibility in HF to work with linear data and convert the image at the end to non-linear by applying an camera.icc profile. (During the process the image should be displayed non-linear, of course :) )

Or maybe it is possible to process the exposure-slider non-linear?

But I think these things should be in the suggestions forum... :wink:

User avatar
Ariel
Posts: 395
Joined: 26.10.2006 06:05

Unread postby Ariel » 30.08.2007 06:37

There are other sliders in the licensed versions (in expert mode) for more/better control of exposure.

firexball
Posts: 21
Joined: 07.07.2007 18:10

Unread postby firexball » 30.08.2007 14:03

Hello!

I think, if Helion could advertise with an full 16bit linear workflow, this should lure/persuade new customers. In some communities the linear workflow is something like the "Holy Grail" of EVB.

I think it shouldn't be too difficult to implement this in HF.
In the Preferences there should be an checkbox to activate the linear workflow (maybe only in combination with DCRaw linear output option) and the possibility to choose an camera.icc profile. The processing of the image should stay the same (all with the linear image), except that before the image is dispayed it should be "married" with the camera profile (or conterted to Gamma 2.2 ?). Maybe this is the most complicated part, but in my eyes this should be nothing different to a sRGB or AdobeRGB conversion!?!

Greetings firexball

edit: The benefit of this would be, that the Problem above should be solved!

lutz
Posts: 155
Joined: 14.04.2006 02:51

Unread postby lutz » 31.08.2007 03:09

Hello Firexball,

can you provide a link to any "real-world" image quality differences achievable by linear processing. I only know of Lightzone claiming such advantages, however they do not provide any good demonstration (or any at all?) of these advantages. Theoretically it sounds very interesting.

thanks
Lutz

firexball
Posts: 21
Joined: 07.07.2007 18:10

Unread postby firexball » 01.09.2007 00:52

Hello!

I just want to answer/comment some of the replys of my post.

to Ariel:
Yes. of course, there are really enough sliders to correct an image in HF. I just wanted to show, that the exposure slider doesn't do what I expected. If I only want to correct the exposure of an image I don't want to use several sliders to do that... if possible!

to Lutz:
Sorry, I just started to learn about linear processing. At the moment I can't imaging that there are really big quality advantages! I only suggested that, because Dan Kozub mentioned, that the exposure compensation in HF is linear (maybe I missunderstood that?).

firexball
Posts: 21
Joined: 07.07.2007 18:10

Unread postby firexball » 02.09.2007 11:41

Hello again!

OK, just to demonstrate that a linear workflow should be an advatage for HF, here the test from above with the following workflow:

1. convert RAW with "dcraw -4 -T -w CRW_XXXX.CRW" (this provides an linear 16bit tiff)
2. open with Helicon Filter and do the exposure correction (I only used the exposure slider)
3. save image as 16bit Tif
4. convert linear Tiff to non-linear Tiff with "tifficc.exe -i camera.icc input.tif ouput.tif"

And these are the results:

Image

As you can see, the results are much better (especially at the +1 image) compared to the HF images in my second post. In my opinion the results are equal to the results of dpp!

Just for completeness some more points I have to mention (for the reason that somebody want to try it on their own:
Because the image has been saved in linear mode, the picture appears very dark in HF. So you can't correct the exposure by looking at the picture. I did it by converting the correct exposured image the same way and looked at the histogram. Then I tried to correct the histogram of the wrong exposured images to look the same. (If the linear workflow is implemented in HF like I described above this should be no problem.)
Second I have to say, that the range of the exposure slider in HF was not enough to correct the -2 image. So I applied +100%, saved the result and opend it again to apply +100% again!

Maybe this test is also an answer to the question of lutz. As you can see, in the in the +1 picture, a non-linear workflow isn't able to correct the image, a liniear workflow is.

Another advantage could be that sharpening should be possible with less halos! But this is only what I red in the internet. I haven't tried this yet!

Greetings firexball

User avatar
Dan Kozub
Posts: 1157
Joined: 24.03.2004 18:14

Unread postby Dan Kozub » 04.09.2007 12:09

Thanks for discussion! We've implemented image processing in linear space (without gamma 2.2 compression). I personally do not see substantial difference but you are welcome to test it. I would appreciate your comments very much.

Please send me an email to dankozub at helicon.com.ua and I will give you a link to to the test version.


Return to “Helicon Filter: Bugs and Support”

Who is online

Users browsing this forum: No registered users and 1 guest