Dual ISO - massive dynamic range improvement (dual_iso.mo)

Started by a1ex, July 16, 2013, 06:33:50 PM

Previous topic - Next topic

1 Member and 1 Guest are viewing this topic.

spider

I have two problems.

First one:

cr2hdr does not work in command line for me with all pictures.


The command "exiftool"(dcraw) is wrong written or could not be found. But both exe are located in the same folder with cr2hdr.exe


Second one:
So I used the barracudaGUI and it works with it, but not with this this picture https://www.dropbox.com/s/y6fybg1i2ta6149/IMG_2524.CR2
Other user reported, that he also can not process it with cr2hdr.

Input file      : IMG_2524.CR2
Camera          : Canon EOS 650D
Full size       : 5280 x 3528
Active area     : 5208 x 3476
White level     : 12500
Black borders   : 72 left, 52 top
Black level     : 1992
ISO pattern     : BdBd RGGB
Interlacing method not supported
ISO blending didn't work


This picture is taken with 650D and FEB 20. build

My OS is Windows 8.1

moonlite

hi!

is there any way to extract the two "exposures" so i can combine them manually, as i do any other HDR?

best woud be if i could get two raw-files exactly as i would have shot two separate images like traditional HDR.

i already tried the --debug-blend -option but all the images i get from that look exactly the same. except the what is supposed to be the bright picture which is exactly the same except the hilights have a pink tint.

any suggestions?

my camera is a 550d and i use a ml version less than a month old

a1ex

Develop the output RAW with two different exposure compensations, and blend them in the classic way.

Next update will have an option to burn some exposure compensation in the DNG file (will compress the highlights and hopefully extract a little more detail from shadows).

Marsu42

I just tried the new 20bit processing on a couple of my dual_iso files. The good news: I don't see any regressions! The bad news: It looks exactly the same as before :-p ... where do I look for changes, i.e. what's the visible benefit of the increased resolution?

a1ex

There's no increased resolution; the difference may be in extreme shadows (if you ever got posterization). The most obvious difference is with my first fake_dual_iso test (which is quite extreme, 6 EV between the two exposures).

If you feel you are close to the 16-bit limit, try --soft-film.

Quick sample (fake dual ISO 6EV):

       

1. 16-bit engine (522eb2e), developed with +10 EV in ufraw-mod, --clip=digital
2. 20-bit engine (b13fcd5), developed with +10 EV in ufraw-mod, --clip=digital
3. 20-bit with --soft-film=4, developed with +6 EV in ufraw-mod, --clip=digital

The --soft-film option compresses the highlights and bakes an exposure compensation in the DNG. With this trick, it uses more bits for shadow detail and you get a brighter image to start with. The baked curve looks like this:



For baking a soft film curve, you need to know the WB in advance, at least roughly (read the multipliers from ufraw, for example). In all the examples I used 4000K/green=0.8, which has the multipliers 1.899, 1, 2.173. Since the default WB for these baked curves is 2,1,2, this fits under the engineering definition of Pi (which is 3), so I didn't bother changing the multipliers for now.

Let's play with it a bit more:
       
4. 20-bit with --soft-film=8, developed with +2 EV in ufraw-mod, --clip=digital
5. 20-bit with --soft-film=9, developed with +1 EV in ufraw-mod, --clip=digital
6. 20-bit with --soft-film=10, developed with +0 EV in ufraw-mod, --clip=digital

Now, let's try to bake the same WB as the one used for developing, to see how much it matters:


7. 20-bit with --soft-film=10 --wb=1.9,1,2.17, developed with +0 EV in ufraw-mod, --clip=digital

Notice the subtle WB change in highlights.

Alright, so we got pretty good highlight/shadow detail now. How does it compare with ufraw-mod's --clip=film?

   
8. 16-bit engine, developed with +10 EV in ufraw-mod, --clip=film
9. 20-bit with --soft-film=4, developed with +6 EV in ufraw-mod, --clip=film

In image 8, notice how ufraw-mod sacrifices 2 shadow bits in order to get better highlight detail (compare with 1). That's a limitation of its 16-bit processing engine. Floating-point raw processors would not suffer from this => consider the highlight compression as a trick to bypass the bit depth limits in either the DNG container or in the raw processor itself.

So, you have noticed this curve is exactly the soft-film curve from ufraw-mod, and now you have the option to bake it in the DNG (you can consider it as a pre-grading). You can bake it completely (images 6,7) or just a part of it (9). Experiment and see what works best.

Experimental binary: cr2hdr-20bit.exe


This example was extreme (I chose it to push the code beyond reasonable limits). In practice, the improvement of the 20-bit engine is probably minor, I'm not sure if it's worth the small speed penalty and increased memory usage. You tell me (if you find differences, just post some examples).

Happy pixel peeping!

Marsu42

Quote from: a1ex on March 06, 2014, 12:09:42 AMIn practice, the improvement of the 20-bit engine is probably minor, I'm not sure if it's worth the small speed penalty and increased memory usage. You tell me (if you find differences, just post some examples).

Right, that explains it :-) ... with increased resolution I meant the internal processing depth, I know there won't be any additional magic done to dual_iso files that actually exchanges the Canon for a Nikon sensor. As for speed, at least on my box cr2hdr is so slow it hardly makes any difference so you may as well keep the new version just to be on the safe side.

hjfilmspeed

Worked great for me didnt seem any slower anyway. I dont know if im taking advantage of the 20bit through lightroom or acr cuz i think they only do 16bit (this was probably discussed but im sure i missed it) but results are still impressive! Well done!

Marsu42

Quote from: hjfilmspeed on March 06, 2014, 04:33:38 PM
I dont know if im taking advantage of the 20bit through lightroom or acr cuz i think they only do 16bit

No matter, 20bit internal processing gives you the high quality, high end feeling that was simply missing from the old and shaggy cr2hdr versions :-p

a1ex

Actually I did encounter a real-world improvement in a picture from what I shot this weekend. It looks pretty much like my example from above, so I've re-converted that picture with --soft-film=4 and got the shadow detail back.

It was just a plain old 100/1600 with deep shadows, exposed with ETTR (nothing fancy). If anyone is interested, I can develop that image with a few different settings to show the differences.

Marsu42

Quote from: a1ex on March 06, 2014, 05:31:50 PM
It was just a plain old 100/1600 with deep shadows, exposed with ETTR (nothing fancy). If anyone is interested, I can develop that image with a few different settings to show the differences.

Well, if it's different from the sample shots above I guess it might be interesting to see where dual_iso and cr2hdr have their limits with the current 16bit output, i.e. what is just barely recoverable and what only shows after applying a pre-processing curve.

hjfilmspeed

I always love me a good pixel peeping session! Again great work! Im forever impressed

naturalsound

This 20bit version with soft-film=2 really solved a severe colour blocking problem I faced in a skyline HDR of Melbourne.
With the old version the sky consisted of huge pink / blue squares. Now it is all smooth.

GREAT :-)

budafilms

Remember the thread about when it's convenient use dual ISO?
I need to re check the possibilities of this module...

Luiz Roberto dos Santos

@a1ex

What edge-detection method cr2hdr use? EECI? I find one very interesting paper, of three koreans, in proposing the new method of edge detection method and the refinement scheme, called "DVBDM" or "Directional Variances Based Demosaicing Method". See here.
The algorithm described is 37,8% more faster than EECI (avarage), and have more 2.95 dB of result on Color PSNR measurement.
I was wondering if it would be possible to adapt something to use in cr2hdr... but unfortunately I am not mathematician to do it.

*only problem of that, which is used is 5x5 CS.

swinxx

Is there a 20bit version for mac? I have a di shot with some trees and the sun and after conversion i have many pink aeras..

norbyg570

I have the same problem like mad77 i can see the scanlines in the CR2 file, but when i start cr2hdr.exe the dng becomes the same like the original raw, without dual iso.

Audionut

Quote from: norbyg570 on March 13, 2014, 12:35:51 PM
I have the same problem like mad77 i can see the scanlines in the CR2 file, but when i start cr2hdr.exe the dng becomes the same like the original raw, without dual iso.

You should provide some samples.

nava

Greetings and Salutations
I see some users have been playing with the 600D to get the Duel ISO for HDR somewhat working.  I did read the original PDF so its a little confusing if this actually works for the 600D?  Or if it will at some point..deciding on a new camera to buy :)


norbyg570

here are some samples:

the RAW CR2 dual ISO


the DNG file from dual ISO


Audionut

By samples, I meant the dual_iso.CR2 file.
Nevermind.


Now open that DNG file in your favorite image editor.  Boost the exposure.  Take note of the noise in those dark areas.
Do you have the same image taken with normal ISO settings?  If you do, open that file in your image editor also, and boost exposure.  Note the difference in noise compared to the dual_iso file?

Now take the dual_iso DNG, and post process (tone map) that file as needed.  Job done!

The reason why dual_iso images are so dark, is that you have captured highlights that are significantly brighter then the midtones/shadows.  So you need to post process them to bring the midtones/shadows higher.

You can do the same with standard ISO also, but note the large difference in noise.

kgv5

I would like to ask about 'custom file prefix' option in dualISO module. It seems that this works only with cr2 for now (and its buggy according to description), is it possible to activate it for movies? I tried and it doesnt affect movie file names. With the new apps and scripts it would be very useful to pick up dualiso files easly and fast from couple houndreds mixed MLVs.
www.pilotmovies.pl   5D Mark III, 6D, 550D

norbyg570

thanks! i will try it this weekend and post the difference

tonybeccar

Quote from: kgv5 on March 14, 2014, 09:25:40 AM
I would like to ask about 'custom file prefix' option in dualISO module. It seems that this works only with cr2 for now (and its buggy according to description), is it possible to activate it for movies? I tried and it doesnt affect movie file names. With the new apps and scripts it would be very useful to pick up dualiso files easly and fast from couple houndreds mixed MLVs.

Indeed! I just implemented dual iso video support in my converter and that would be neat! Though I think there IS a way of detecting automatically with metadata but that would be MUCH easier..

arturochu

is anyone having this kind of flicker?

https://vimeo.com/89151079

any thoughts on how to remove it?
Chu