PinkDotRemover tool 650D

Started by foorgol, June 15, 2013, 08:51:57 AM

Previous topic - Next topic

0 Members and 1 Guest are viewing this topic.

gary2013



maxotics

They look the same.  Both have pink dots.

gary2013

here is one dng where pdr just worked. i just shot this at 1280 non crop 24 fps. i applied the raw file to pdr ver 8

https://www.transferbigfiles.com/d02e118a-045a-4ffe-9fbb-5388a5c9ee0f/hPUFDyJ_cEnxKWMVcwqgBQ2

the other files i did at 1728 30 fps, if i remember correctly.

gary

gary2013

Quote from: maxotics on October 02, 2013, 10:33:15 PM
They look the same.  Both have pink dots.
correct. pdr did nothing to fix it. i also tried both versions of pdr with the same results

a1ex

Interesting that a simple 3x3 chroma smoothing like the one from cr2hdr seems to fix the pink dots without any prior info about them.

http://a1ex.magiclantern.fm/bleeding-edge/1280x434_FocusDotsExampleLightTable-cs3x3.DNG

gary2013

here is one more dng file where pdr did not work. this is from the raw file i shot earlier today. the raw file is fresh and i ran it thru pdr successfully. i then ran raw2dng and here is one frame form that sequence that shows pdr did not work.

https://www.transferbigfiles.com/b23d6469-3344-4956-8ab0-2034652e14ce/IDQnDV3dn3_zwJT2oaXFRg2

gary
the only thing different in mine were the earlier one being shot at 1728 non crop and the second one where pdr worked being 1280 non crop


Rewind

Quote from: a1ex on October 02, 2013, 10:44:39 PM
Interesting that a simple 3x3 chroma smoothing like the one from cr2hdr seems to fix the pink dots without any prior info about them.

A1ex, can you apply the chroma smoothing to this dng, so I can compare results with PDR?

a1ex

Here you go: http://a1ex.magiclantern.fm/bleeding-edge/rewind.dng

Can you generate the jpeg for pixel-peeping?

Edit: here are some experimental versions of raw2dng that perform chroma smoothing:

http://acoutts.com/a1ex/raw2dng_cs3x3.exe
http://acoutts.com/a1ex/raw2dng_cs5x5.exe

They may help with color aliasing too, so try them both and see which gives the best results.

Rewind

Quote from: a1ex on October 02, 2013, 11:08:56 PM
Can you generate the jpeg for pixel-peeping?

Two jpegs:
by Chroma smoothing
by PDR

Color fringing is more pronounced in first picture. This is quite expected result i think.

maxotics

Looks good!  However, I would think that it would forfeit 30% of your color information--going down that slippery H.264 slope.  Will do more tests...  Thanks!  (BTW, besides g3ggo, has anyone ported any of these routines/libraries to C#?)

a1ex


a1ex

Pixel peepers: which one do you like best and why?

original dng png
3x3 (first version) dng png
3x3_v2 dng png
3x3_median4 dng png
3x3_ahdlike dng png

Rewind

Quote from: a1ex on October 03, 2013, 12:07:38 PM
Pixel peepers: which one do you like best and why?
ahdlike has less color artifacts

spider

For me the last one looks best, because there is nearly no color aliasing.

a1ex


Rewind

by the way, a1ex, since green channel stays untouched (focusing pixels are red and blue only), is it possible to calculate chroma on R and B channels only? This way we will preserve more detail.

I mean, somehow make the AHD-like calculations on R&B channels only

a1ex

I'm not touching the green channel.

spider


a1ex

Small update: noticed that "ahdlike" affected noise structure in Gary's shot (which is extremely underexposed), so I've added a threshold below which it no longer tries to choose the interpolation direction (it averages both).

raw2dng_cs3x3_ahdlike_noise.exe
raw2dng_cs5x5_ahdlike_noise.exe (for really noisy shots)

For regular shots, the noise threshold should not make any difference (if it does, post some test shots that show the difference). For noisy shots, the noise should be whiter (less correlated, easier to clean).

Rewind

Unfortunately, this method produces pretty obvious artifacting in small detail, as well as expectable overall color detail loss.

a1ex

Then, extract a list of pink dots and only interpolate these points. The list I've received was incomplete.

Rewind

Quote from: a1ex on October 03, 2013, 01:45:58 PM
Then, extract a list of pink dots and only interpolate these points. The list I've received was incomplete.

This is exactly what we are doing in PDR, but to try this AHD method, I need it's math description. Would you mind to share one?

a1ex


Rewind