Messages

Can we play with the raw files? This seems a very good test for fine-tuning my HDR script.

@dubzeebass: for banding, you started the video with a very underexposed image. Banding correction is computed from the first frame only, so if you start from the sky, it should be a lot better.

You may try something really extreme now... maybe a photo from a cave outside? or maybe something like this or this?

In raw you are getting only full stops (you can't get digital gain). Make sure you expose to the right.

This update should be pretty close to the ideal auto ISO option - for my needs.

https://bitbucket.org/hudson/magic-lantern/commits/373d8b5b0e69

You expose for highlights, it takes care of the shadows. It's pretty much impossible to do the other way, because you can't figure out how much a picture is overexposed if you look at the histogram.

Did it melt?

see http://www.magiclantern.fm/forum/index.php?topic=1573.0

The old converter from here was different from the one used for the previous set (the shadow artifacts are no longer there). Was it median6 (the very first version)?

I don't see any quality loss with the new converter. If anything, the sharpness was set too high when postprocessing the picture.

I'd also try these DNGs with this script. Would be nice if you could upload one, maybe when you come back.

Can you do now a comparison between ISO 100 and ISO 100/1600?

The option from FPS override is for H.264 (and it's not syncing either).

I first tried 100 on the camera and 800 on ML. But get better results if I do it the other way around. Is that the way this is supposed to work? Doing it the 800/100 way gives me better histogram accuracy and highlight /shadow feedback while recording.

Is there any difference in the end result? Can you post a comparison?

Small converter update:

- fix crashes in cr2hdr with some unusual pics
- fix red "zebras"
- nicer highlight handling (aliasing artifacts should be blurred)
- minor tweaks

Yes, you see it in the bbDD pattern. For me, only that one clip from Luke Neumann is crawling; everything recorded by me, and the Batman clip (also from Luke Neumann) are stationary. The clip from EOSHD is also stationary.

I'm not yet sure when crawling occurs; it appeared in one test shot from Luke Neumann, but it didn't appear in all of my other test shots...

Maybe it has to do with FPS, PAL/NTSC or something like that.

It was not moved anywhere - this feature was removed because sound was out of sync.

@JackDaniel412: record 0.5 seconds of video showing the problem, and upload the RAW file. The DNG is processed and doesn't help.

@stevefal: this patch should alternate ISO exactly the way you suggested: http://acoutts.com/a1ex/alt_iso.patch

You can play with it, see if you get improvements, and I'll probably add it as an option in menu.

You have the full source code, feel free to modify it. I don't see the point for this.

See the post above.

For now, disable the modules, enable CONFIG_PICOC and compile it yourself. Or, download some older nightly (not sure when they were disabled).

It's on the todo list, but it will take a while. Just compare reactions for scripting vs raw video stuff.

@Danne: try processing the same shots with the latest cr2hdr. You should get much cleaner shadows (so I shouldn't have to read the text to know which is which )

In particular, the black spots should no longer be there.

FYI: http://www.chiark.greenend.org.uk/~sgtatham/bugs.html

You should not expect us providing a polished product that you can just consume. This is a community effort. Grab your C compiler, fix it and submit a patch.

I've left this one unfixed on purpose - it's a great opportunity to get your hands dirty and learn something.

Continue here: http://www.magiclantern.fm/forum/index.php?topic=7140.0

Is a 0.1 EV difference that important?!

Or are you taking one picture per hour?

I'm afraid what you are requesting is not possible (unless you do the reverse engineering yourself).

I get error 1 in photo & LV, reg not 0?

From the dumps, it's register 3, right? (I didn't use the original dump format, but some custom hooks)

Try overriding it manually (just change that memory address so it reads 0x3 instead of 0x33, for example).

/* 00:00:05.898461  */     03 0000 40450E08
/* 00:00:25.234033  */     03 0033 40450E3E
/* 00:00:55.186093  */     03 0044 40450E50
/* 00:01:23.931307  */     03 0055 40450E62
/* 00:01:39.858203  */     03 0077 40450E74

This one looks really good.

Code Select Expand


        ?_CMOS_ISO_START = 0x40450E08;
        ?_CMOS_ISO_COUNT = at least 7; // from ISO 100 to 6400
        ?_CMOS_ISO_SIZE  =  18;

        CMOS_ISO_BITS = 4;
        CMOS_FLAG_BITS = 0;
        CMOS_EXPECTED_FLAG = 0;


If it works, the theory about 8-channel readout is invalidated (the 6D has 4 channels afaik).

For some reason, dcraw was writing a pixel value of 65506 (come on, for 14-bit data, the range is 0-16383...). This also breaks my calculations (even the robust statistics part). The image is read correctly though.

I've converted the shot by hacking the source code a bit, but before providing a fix, I want to understand exactly what's going on.

http://acoutts.com/a1ex/IMGL9830.DNG
http://acoutts.com/a1ex/IMGL9830.jpg

make raw2dng.exe