Project

General

Profile

Camera Support #10655

pentax k-s1 - add raw support

Added by Giuseppe Savo about 4 years ago. Updated almost 4 years ago.

Status:
Incomplete
Priority:
Low
Assignee:
-
Category:
-
Target version:
-
Start date:
10/04/2015
Due date:
% Done:

20%

Estimated time:
Affected Version:
1.6.8

Description

Darktable doesn't support .pef raw files from pentax k-s1.
I'm uploading one of these files. The sensor on k-s1 sholud be the same present in ks2, so the pef files should be similar...

I uploaded a pef file here:
https://www.dropbox.com/s/1zwztpiatx8l773/FOTO0468.PEF?dl=0

Associated revisions

Revision d852c983 (diff)
Added by Johannes Hanika about 4 years ago

noise profile: ricoh pentax k-s1 by guiseppe refs #10655

History

#1 Updated by Pedro Côrte-Real about 4 years ago

  • % Done changed from 0 to 50
  • Status changed from New to In Progress
  • Category deleted (Darkroom)

Basic support has now been added in this commit:

https://github.com/darktable-org/darktable/commit/8939722c0199a690c7c300fcb6d47349b87faf7b

If you provide files for WB presets and noise profile we can add those too. Thanks for the same

#2 Updated by Giuseppe Savo about 4 years ago

I'll upload ASAP the required images for noise profile and WB presets.
I don't have a colorchecker, but I can ask to a friend if I can borrow it from him.

#3 Updated by Pedro Côrte-Real about 4 years ago

The color target is the least important in my opinion. The standard matrices are usually fine and the user supplied matrices are easy to get subtly wrong which comes to bite you later (e.g., we've seen strange hues of blue in some pictures from a wonky matrix).

#4 Updated by Giuseppe Savo about 4 years ago

Hi!
I tried to make some shots to create a profile for the noise.
After 2 different series of shots, I obtained results similar, but different. In the first shots I used a warm light, during the second session the same led light was helped with a "cool" ribbon of leds.
Is it possible that this difference in temperature of light can bring to different profiles?
And now... which is better? Are the profiles so different to bring to better/worse results?

I uploaded preset and presets2 corresponding to the different situations (presets cool light, presets2 warm light). The presets "cool" is not complete: the second or the third shot gave an error at beginning, so I used these clicks just to have something to compare.

https://www.dropbox.com/sh/y9u0yr51s74ygah/AAC-MnmVDfglK4czqbCgjTEia?dl=0

EDIT: ok, please, don't use this!!! take the other one.

#5 Updated by Pedro Côrte-Real about 4 years ago

Please also submit at least one sample file to rawsamples.ch

#6 Updated by Giuseppe Savo about 4 years ago

Hi!
Today I realized a though: the ks-1, similarly to other pentax cameras, can pass trough the ISO by 1/3 EV or 1/2 EV steps.
So I created another sequence, introducing both the series:

https://www.dropbox.com/sh/r58e3euur7fg7ya/AAB7rlqmPKj6X2VePjRw-ToXa?dl=0

Now I have a question (another one): changing temperature of light, the values of blue and red are quite different. Is this an issue?
I'll charge various dng on rawsamples.ch, probably tomorrow (today the ftp server seems not working...). My images are FOTO0xxx.DNG.

#7 Updated by Pedro Côrte-Real about 4 years ago

I'll charge various dng on rawsamples.ch, probably tomorrow.

Please don't submit converted DNGs if that's what you mean. A lot of the Pentax cameras produce both DNG and PEF so if that's what you mean please submit one of each. Nicely exposed images with good color are preferred but anything you have on hand will be fine.

#8 Updated by Giuseppe Savo about 4 years ago

Pedro Côrte-Real wrote:

I'll charge various dng on rawsamples.ch, probably tomorrow.

Please don't submit converted DNGs if that's what you mean. A lot of the Pentax cameras produce both DNG and PEF so if that's what you mean please submit one of each. Nicely exposed images with good color are preferred but anything you have on hand will be fine.

Ok. tomorrow I'll make some "good" images at various iso, and I'll submit the dng's. I tried to charge dng files on rawsamples, but the ftp server was "busy".

#9 Updated by Pedro Côrte-Real about 4 years ago

Ok. tomorrow I'll make some "good" images at various iso, and I'll submit the dng's. I tried to charge dng files on rawsamples, but the ftp server was "busy".

Cool. For rawsamples having the several iso isn't as needed. Having a PEF as well would be good though.

#10 Updated by Giuseppe Savo about 4 years ago

ok, I uploaded dng and pef files on rawsamples.

Please, consider the second set of noise data instead of the first I uploaded. :)

I'll pass to the colors asap.

#11 Updated by Pedro Côrte-Real about 4 years ago

  • % Done changed from 50 to 20
  • Status changed from In Progress to Incomplete

The noise profile has landed already, the only relevant thing now are the wb presets. You can just shoot whatever (can be a black image) with each of the profiles in the camera and label each of the files so we know which profile is which.

#12 Updated by Giuseppe Savo about 4 years ago

Hi Pedro,
please, if possible, change the noise profile. The one I uploaded as first doesn't appear "good". Yesterday and 2 days ago I tried to work on images with that profile, and the results weren't satisfactory...
...actually neither "decent".

about the WB, isn't it necessary a colorchecker? If not, tomorrow I'll produce some images. If it's better a colorchecker, I'll have one available on saturday. :)

#13 Updated by Pedro Côrte-Real about 4 years ago

Change to what? Johannes was the one that evaluated it and merged it so he apparently thought it was decent. Do you have a better one to use?

#14 Updated by Giuseppe Savo about 4 years ago

I think that this

https://www.dropbox.com/sh/r58e3euur7fg7ya/AAB7rlqmPKj6X2VePjRw-ToXa?dl=0

is better.

furthermore, in this series I took an image for every step (1/3 or 1/2 EV), respect to the other profile where are present only images stepped by 1/3 EV.

EDIT: I tried the profiled noise reduction, and in my opinion the channel of red is too "strong", and the channel of green is too "weak".

#15 Updated by Pedro Côrte-Real almost 4 years ago

Giuseppe Savo wrote:

I think that this

https://www.dropbox.com/sh/r58e3euur7fg7ya/AAB7rlqmPKj6X2VePjRw-ToXa?dl=0

is better.

I've looked at the fits and they don't look very close so I think we better just keep the ones we have already. The camera may just not fit our noise model very well.

furthermore, in this series I took an image for every step (1/3 or 1/2 EV), respect to the other profile where are present only images stepped by 1/3 EV.

This shouldn't really matter as we interpolate between iso levels as needed.

EDIT: I tried the profiled noise reduction, and in my opinion the channel of red is too "strong", and the channel of green is too "weak".

Which ones did you test, the ones we have already merged?

#16 Updated by Giuseppe Savo almost 4 years ago

Hi Pedro,
sorry for the delay, I was without the camera for some weeks.

I'm not satisfied of profiles I obtained: the correction is too heavy, and there is a sort of "green noise" after the correction.
I tried using the parameters from sony SLT A58, a camera with the same sensor, obtaining better results, but the jpg's produced in camera are more detailed and less grainy...

Also available in: Atom PDF

Go to top