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

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

Previous topic - Next topic

0 Members and 6 Guests are viewing this topic.

tron

Something else: Temperature changed from 3100 (tint +3) to 4800 (tint -9) after conversion

tron

I will do the same experiment tonight with long exposure NR  and I will let you know...

a1ex

Color matrix is hardcoded from 5D3. I thought exiftool will take care of it, but this info is not in the CR2 (it's hardcoded in dcraw).

Don't know an easy way to fix it. Maybe IliasG?

arrinkiiii


1%



Can anything be done for bad pixels like this?

Already post converted by raw2dng... originals are raw files like 500Mb-1GB

http://www.filedropper.com/m28-2304-000333

http://www.filedropper.com/m28-1958-000178

arrinkiiii


arrinkiiii


a1ex

QuoteCan anything be done for bad pixels like this?

Try extracting the interlaced DNGs with an older raw2dng (or delete the HDR code from it). Then, decompose them by enabling the debug flags in cr2hdr (those files with "dark.dng", "bright.dng", "fullres.dng" and so on).

If the hot pixels are not isolated, they are not corrected because they may be mistaken for aliasing. So, a hot pixel is only corrected if it's clearly brighter than its neighbours.

edit: I've just removed the old processing code from raw2dng, because:

- raw format is near end of life
- quality is much better with cr2hdr
- the cr2hdr workflow isn't that hard (extra step: select all, drag and drop)
- the algorithm is still improving, so it's easier to maintain a single way for doing things

1%

The hot pixels show up as purple instead of red. It never corrected them (dual or not), esp the ones from FPS override. not sure why, they don't move and are very obvious. Also the color noise/banding in the shadows.

I'll have to try to get all of the files separated and see what is produced.

a1ex

There's no temporal analysis in the hot pixel algorithm, so they may not be that obvious mathematically.

1%

I'll have to find the worst offenders and break them up/post them. The noise is easier to get rid of than the hot pixels afterwards.

rufustfirefly

I've been building Linux and Windows binaries for cr2hdr, as the code seems to be changing pretty often. They're available here, if anyone wants them, tagged with the date and Mercurial revision:

https://bitbucket.org/rufustfirefly/magic-lantern/downloads
EOS 5Dmk3 | EOS 40D
EF 70-200mm f/2.8 IS II ; 50mm f/1.4 ; 24-104mm kit lens ; 135mm f/2.8 SF | FD 28mm f/2.0 ; 55mm f/1.2 ; 300mm f/4 | Nikkor 55mm f/1.2 | Other : Rokinon Cine 14mm, 24mm, 35mm, 85mm ; Juplen M42 135mm f/2.8 ; ... etc ...
http://jbuchbinder.com/

a59192

Quote from: rufustfirefly on August 29, 2013, 07:43:26 PM
I've been building Linux and Windows binaries for cr2hdr, as the code seems to be changing pretty often. They're available here, if anyone wants them, tagged with the date and Mercurial revision:

https://bitbucket.org/rufustfirefly/magic-lantern/downloads

Thanks! rufustfirefly sharing, very helpful to me.  :D

swinxx

tried to convert some video dngs (son of a batch) in vmware windows 7 with the latest cr2hrd but got an error..
so what can i do now?

black subtrackt didn´t work
bright dark detection didnt work
iso blending didnt work..

greets. sw


Francis

Quote from: swinxx on August 29, 2013, 11:24:48 PM
tried to convert some video dngs (son of a batch) in vmware windows 7 with the latest cr2hrd but got an error..
so what can i do now?

black subtrackt didn´t work
bright dark detection didnt work
iso blending didnt work..

greets. sw

Sounds like you are lacking dcraw.

a1ex

That would print dcraw didn't work or something like this.

swinxx

i´m sorry alex.

i just recorded a dual iso enabled video raw file with the latest audionaut 5dmk3 build , then converted it with son of a batch onto my mac 10.8.4
then i had a directory with all the dng files onto my main hdd. with a vmware windows 7 machine i copied all the files into my desktop into a cr2 directory and then i selected those files (10 of them) and dragged them onto the latest cr2hrd.exe file..
with the cr2 photos it works..

thx. and sorry for my bad description, sw

a1ex

Did you read my link? I know it's long, but... I have no idea what's going on. At least upload one DNG ;)

swinxx

dear alex.
thx, i wil upload when i am at home.. (in a few hours.)

greetz.

tron

@a1ex: Night experiment with same theme, exposure, lighting, shutter and aperture with Long Exposure NR set to ON.

No big changes. Strange patches continued to exist. Same image (no dual iso) with NR set to OFF at ISO 1600 at 10sec has very few and unobtrusive hot (of various colors) pixels.

Can we have a choice whether to activate white and/or black pixel fixing? 

P.S I can send you a CR2 of course but it is VERY similar.

swinxx

here is the link to the dng sequence, and to the raw file.
i had no success to convert it to a normal dng seuquence because of the error (seen in the screenshot. ) thx

https://www.dropbox.com/sh/wzh6q2994ve7i6d/kKL5Whmuln

a1ex

Okay, will fix.

Meanwhile, try a more aggressive setting (not 100/200).

Karmaschinken

Same thing here, can´t cr2hdr the dngs extracted from raw vid. Thanks a lot for fixing, Alex!  :-*

Karmaschinken

Sorry, I suppose this is my mistake, obviously the DUAL ISO is not yet implemented for 5D2, right? Can´t wait to have it on my cam soon.