DualISO, Auto-ETTR and lots and lots of pink

Started by RobFlanders, March 27, 2015, 05:13:23 PM

Previous topic - Next topic

0 Members and 1 Guest are viewing this topic.

RobFlanders

Hi all,

I decided to experiment with the DualISO module to take a timelapse of a sunset - so I combined it with Auto-ETTR.

Normally my workflow is:

Import .CR2's
Rename
use UFRAW-BATCH to interpret Auto-ETTR
use FFMPEG to timelapse.


For the DualISO pictures when I use UFRAW-BATCH the pictures come out very grainy and very pink. I have attempted to use cr2hdr.exe to pre-process the DualISO before using UFRAW, but when I try and run UFRAW on the new dng files UFRAW can't find it as it is looking for the original .CR2.

Can anyone explain what my workflow should be in this situation?

http://i.imgur.com/AuVnZSE.jpg

a1ex

Quote from: RobFlanders on March 27, 2015, 05:13:23 PM
when I try and run UFRAW on the new dng files UFRAW can't find it as it is looking for the original .CR2.

That's the issue. I think you could try replacing CR2 to DNG in all *.ufraw files.

Alternative: discard the *.ufraw files (move them in another dir) and try www.magiclantern.fm/forum/index.php?topic=7022
(it's not meant for timelapse, but if you completely disable highlight and shadow recovery as explained here, it may work just fine).

RobFlanders

Thanks for the help.

I changed the XML in the UFRAW file to look for a .DNG source file, rather than .CR2, and it interpreted fine - but it is now quite grainy....

http://imgur.com/2zc6QrQ

a1ex



a1ex


RobFlanders

Added a new link. Not sure why the old one didn't work.

a1ex

With dual ISO, you get only ~14 stops of DR, not 20 :P

Set the SNR limits to some higher value; it will clip some more highlights, but shadows should be much better.

garry23

I thought I would throw in some of my experience here, as I'm sitting down having an early Saturday brunch in sunny New Mexico.

In my experience Dual-ISO runs into difficulty when you don't fully appreciate the DR of the scene. As Alex has stated, Dual gives you an extra few stops of DR cover, but the quality of your final data is highly dependent on the quality of the final image you are seeking and over what DR you want to cover.

I tend to adopt a three layered strategy for exposure capture, ie this will be complicated by other needs, such as focus stacking.

The key, first step, is to evaluate the scene, ie its DR over the areas you are interested in capturing. I personally use a 1 degree spot meter, but fall back to my camera's spot if I don't have my external meter with me. BTW if you are using a long lens you will approach a 1 deg spot, but with a wide angle this, obviously, will not be the case.

Once I understand the scene's DR I decide on my capture strategy.

For a low DR, ie covered by a single, non-Dual, I will simply use ML's A-ETTR.

For a scene with a 'medium' DR, ie covered by Dual, I will use A-ETTR with dual. I very rarely use the dual extra S/N features, as these take control away from me.

Finally, and always on a tripod, I use ML auto bracketing in this way. I first meter the darkest area where I wish to capture detail and use this as my base exposure. Then I'm guaranteed that auto bracketing algorithmics will capture the best bracket set. What you don't want to do, IMHO, is to start auto bracketing 'in the middle'.

I hope the above is of some value to those experimenting with ML based exposure capture.

PS I also posted a few reflections here: http://photography.grayheron.net/2015/03/know-your-basics.html

RobFlanders

Thanks for the info guys, much appreciated.

I will think before firing next time.

RobFlanders

The final post-processed video (I haven't done anything except CR2HDR and UFRAW, then used ffmpeg to timelapse). There's a very obvious jump where ML has turned off DualISO towards the end of the video.

https://youtu.be/tiY57UtTQ1Q