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

[mk11174]

Can someone please add the 550D diff code to the commit?

Code: [Select]

diff -r ee7c743dfb14 modules/dual_iso/dual_iso.c
--- a/modules/dual_iso/dual_iso.c Thu Aug 08 23:03:02 2013 +0200
+++ b/modules/dual_iso/dual_iso.c Wed Aug 28 01:44:59 2013 -0400
@@ -91,6 +91,7 @@
 static int is_50d = 0;

 static int is_6d = 0;

 static int is_500d = 0;

+static int is_550d = 0;

 static int is_600d = 0;

 static int is_700d = 0;

 

@@ -742,6 +743,29 @@
         CMOS_FLAG_BITS = 3;

         CMOS_EXPECTED_FLAG = 0;

     }

+   else  if (   streq(camera_model_short, "550D")       )

+    { 

+               

+        FRAME_CMOS_ISO_START = 0x40695494; // CMOS register 0000 - for LiveView, ISO 100 (check in movie mode, not photo!)

+        FRAME_CMOS_ISO_COUNT =          6; // from ISO 100 to 3200

+        FRAME_CMOS_ISO_SIZE  =         30; // distance between ISO 100 and ISO 200 addresses, in bytes

+

+//  00 0000 406941E4  = 100

+//  00 0024 406941F6  = 200

+//  00 0048 40694208  = 400

+//  00 006C 4069421A  = 800

+//  00 0090 4069422C  = 1600

+//  00 00B4 4069423E  = 3200

+

+        PHOTO_CMOS_ISO_START = 0x406941E4; // CMOS register 0000 - for photo mode, ISO 100

+        PHOTO_CMOS_ISO_COUNT =          6; // from ISO 100 to 3200

+        PHOTO_CMOS_ISO_SIZE  =         18; // distance between ISO 100 and ISO 200 addresses, in bytes

+

+        CMOS_ISO_BITS = 3;

+        CMOS_FLAG_BITS = 2;

+        CMOS_EXPECTED_FLAG = 0;

+                is_550d = 1;   

+    }

     else if (streq(camera_model_short, "600D"))

     { 

         /*



[Audionut]

Can someone please add the 550D diff code to the commit?

Pull request?

[mk11174]

Pull request?

Yes please, I dont know how to do that, sorry. 

[Audionut]

Yes please, I dont know how to do that, sorry. 

It's very easy for small lines of code like this.  Give me a few mins and I'll do a quick tut.
It can all be done via web browser 

[a.d.]

@Oedipax
you may want to try my OSX release with gui?

[mk11174]

It's very easy for small lines of code like this.  Give me a few mins and I'll do a quick tut.
It can all be done via web browser 

cool, sounds good 

[Oedipax]

@Oedipax
you may want to try my OSX release with gui?

Thank you! Didn't know anyone was working on this.

[Audionut]

cool, sounds good 

Here you go

http://www.magiclantern.fm/forum/index.php?topic=7940.msg70958#msg70958

a1ex has already merged your code that I used to create the tutorial.
https://bitbucket.org/hudson/magic-lantern/commits/2a3faad36479e6bb576cd03ba10ec2a6649fe803

[tron]

@tron: this time you are oversharpening the picture. I don't see anything bad in my jpeg.

I hadn't  changed the sharpening in any case. I had just restored WB to auto (instead of as shot) to make the image realistic (to match the wall color).

[mk11174]

Here you go

http://www.magiclantern.fm/forum/index.php?topic=7940.msg70958#msg70958

a1ex has already merged your code that I used to create the tutorial.
https://bitbucket.org/hudson/magic-lantern/commits/2a3faad36479e6bb576cd03ba10ec2a6649fe803

Ok, but still, thanks for the tut, at least it will help in future. 

[f3000]

Maybe the hot dot remove algorithm is a bit aggressive and do harm to the shadow detail，can this be improved？

[pavelpp]

Is dual ISO already supported in raw2dng GUI app for Mac?

[robinlee]

During conversion, why the exiftool always say didn't work when I have them in every folder for conversion?

 

[glubber]

During conversion, why the exiftool always say didn't work when I have them in every folder for conversion?

 

In my case (Win7-64bit) the exiftool worked after doing the convertion in a rootfolder ("C:\DualIso" for example instead of "C:\Pictures\2013\2013-08-28").

[X-RAY]

Wow, the new CR2HDR is very nice for me.
I had one problematic file with magenta patterns in the shadows, but with the new version they're totally gone.
Not just the magenta but also those patterns in the dark. At this point it should be obvious, that we have to save the original CR2-Files for the future.

For demonstration.
The CR2-file:
https://dl.dropboxusercontent.com/u/35226956/DualIso/DUAL4390.CR2

This was processed with the old version (Shadows +100)
https://dl.dropboxusercontent.com/u/35226956/DualIso/DUAL4390_CR2HDR_OLD.jpg

This with the new version (same Settings)
https://dl.dropboxusercontent.com/u/35226956/DualIso/DUAL4390_CR2HDR_08_28.jpg

@ Alex
But there is a small issue. Is there a difference in the embedded Camera Profile? With the old CR2HDR it was "Adobe Standard" and with the new one "Embedded". Did you include something there?
The thing is...do you see this strange looking sun-corona? On the Adobe Standard-Profile this looked perfectly normal and on the embedded profile thats a little ugly. ;-)
But thats just a minor issue.
Nice work!

[a1ex]

I don't touch the exif info, except from copying it from the CR2. Try adding (or deleting) exiftool, to see if it makes any difference.

I also don't see anything strange in the sun, other than a tiny brightness change...

The image is quite oversharpened for my taste.

[tron]

@alex: it is not oversharpened.

https://www.dropbox.com/s/cp954uq8f3reg0r/dual_iso_28-8-0810AM.JPG

I have used Adobe's defaults. You can see them in the screenshot.

I have used the latest cr2hdr (28-8-0810AM) but all the latest look alike.

[a1ex]

Updated cr2hdr again:

- small differences in shadow aliasing and hot pixel handling
  (7D test chart perfectly clean, 5D2 chart - which was more underexposed - almost there)
- should fix exiftool in long folders
- a little less aggressive with nonlinear corrections (not visible on my test shots, but should be less likely to get artifacts in tricky cases)

Before/after, 7D chart:


Before/after, 5D2 chart:


(all compared with the first update from yesterday)

I chose not to fully correct the 5D2 chart (even if it's possible), because in very deep shadows, noise is easily mistaken for detail (and when it happens, it looks very ugly).

@tron: Adobe defaults have too much sharpening, look it up in the raw video section.

[tron]

@a1ex:

https://www.dropbox.com/s/e9yfrvk1odyen41/dual_iso_28-8-0810AM-no-sharpening.JPG
https://www.dropbox.com/s/2ss9i46tb9touf5/dual_iso_28-8-722PM-no-sharpening.JPG

The above are with sharpening set to zero. You can see from the name the version of cr2hdr that was used.

Although Sharpening cannot always be as low as that I'd like your opinion on the above (regarding artifacts)

P.S Is that Canikon instead of the real camera name necessary ? Or you are using  it temporarily to distinguish between versions of the cr2hdr?

[a1ex]

Here's how it looks in ufraw. I don't see anything bad, a little noise is normal.

http://acoutts.com/a1ex/img0020test.png

Here, the camera name for your shot is Canon EOS 5D Mark II. Try changing it from exiftool and see if it makes a difference.

[tron]

Here's how it looks in ufraw. I don't see anything bad, a little noise is normal.

http://acoutts.com/a1ex/img0020test.png

Here, the camera name for your shot is Canon EOS 5D Mark II. Try changing it from exiftool and see if it makes a difference.

It's the "Unique Camera Model" that is added to dng that has value "Canikon". It is displayed by Adobe Camera Raw as you can see in the screenshot that corresponds to the latest cr2hdr.exe

[a1ex]

OK, change it and report.

exiftool -UniqueCameraModel="Nikanon" foo.dng

[tron]

OK, change it and report.

exiftool -UniqueCameraModel="Nikanon" foo.dng

Yes it was changed in ACR 8.1 It says: Camera Raw 8.1 - Nikanon

[a1ex]

Change it to the real thing, see if it makes any difference in output.

[tron]

Change it to the real thing, see if it makes any difference in output.

You mean to set it to Canon Eos 5D Mark II

?? Actually it seems it has been added. So I was thinking to delete it!

But if you need another experiment I can set the value that is displayed by ACR with the original .CR2

EDIT: Done it displays the new value.

I delete it and it displayed: Digital Negative

I have saved the .dng however (no problem)

Why is this happening?