Capture One and Darktable for low light Sony RAW

Hi everybody,

I am not very experienced in PP, so maybe You can help me a bit... For most of my Raw processing, I use Darktable. Recently, however, I noticed Capture One showing less Shot noise with high ISO values and sharper images. The Camera is an A6000 with ARW files.

I would prefer Darktable featurewise (masks, linux compatibility) and the UI, but the end result is the most important thing, so I tried to manually mimic the denoising and sharpening, with little success.

So my questions are:

Does someone know what adjustments C1 does automatically to images after importing?

Do You find significant differences in the camera profiles the softwares use, that could explain that? (as profile data is where the additional funds of commercial development can be put to)

Is there perceivable influence of the RAW Algorythm, or does it all depend on the parameters?

I added a snapshot where the noise difference becomes apparent (stronger denoise in DT costs sharpness for this one). We unexpectedly busted some cute haythieves at a farm in northern Sweden, and did not have much time, so I took a less than perfect picture with the Sigma 60mm f/2.8 ART. Since this is mainly about the noise, I did not match the color profiles in the softwares, so the DT example is a bit more saturated.


Darktable (profiled denoise)


Capture One Express for Sony

Your darktable conversion contains chroma noise which you haven't addressed. Create another instance of the profiled denoise module using the wavelets mode and the color blend(ing) mode. See this post for screenshots.

Also, consider applying (large radius) sharpening only to those areas that really need it, so use masking in your sharpening module (or don't sharpen at all).

A J Fresnel said:

Does someone know what adjustments C1 does automatically to images after importing?

Click to expand...

Phase One create a noise and sharpening profile for every camera /ISO as part of the camera profile, so its not really making any adjustments after importing, but applying them during import.

You can then edit the noise and sharpening to suit, if you don't like Phase One's defaults.

I assume Darktable, is simply not providing the same level of "hand holding" and expecting more input from the user to create their own defaults for each camera/iso combination.

Beyond that, I can't help as I don't use Darktable.

Cheers,

Graham

myotisone said:

A J Fresnel said:

Does someone know what adjustments C1 does automatically to images after importing?

Click to expand...

Phase One create a noise and sharpening profile for every camera /ISO as part of the camera profile, so its not really making any adjustments after importing, but applying them during import.

You can then edit the noise and sharpening to suit, if you don't like Phase One's defaults.

I assume Darktable, is simply not providing the same level of "hand holding" and expecting more input from the user to create their own defaults for each camera/iso combination.

Click to expand...

The basic camera support in darktable includes an input camera profile (ICC, linear, matrix), a generic base curve, camera-specific White Balance presets, and noise reduction profiles. There is also access to LensFun-based collection of lens corrections. Non-optimized, generic input sharpening is added at default but the user needs to activate things like noise reduction and preferably come up with one's own presets for those. Like in Capture One, you can create presets for each of the modules and come up with your own processing styles which can be applied to the entire batch of images on import.

sankos said:

The basic camera support in darktable includes an input camera profile (ICC, linear, matrix), a generic base curve, camera-specific White Balance presets, and noise reduction profiles. There is also access to LensFun-based collection of lens corrections. Non-optimized, generic input sharpening is added at default but the user needs to activate things like noise reduction and preferably come up with one's own presets for those. Like in Capture One, you can create presets for each of the modules and come up with your own processing styles which can be applied to the entire batch of images on import.

Click to expand...

So, are you saying that if you activate it, Darktable will provide optimised noise reduction and sharpening for a specific camera and ISO, as C1 does by default (optimised in terms of what Phase 1 think it means of course)

No reasons why it shouldn't, I was just trying to explain why it appeared C1 was doing a better job.

Cheers,

Graham

myotisone said:

sankos said:

The basic camera support in darktable includes an input camera profile (ICC, linear, matrix), a generic base curve, camera-specific White Balance presets, and noise reduction profiles. There is also access to LensFun-based collection of lens corrections. Non-optimized, generic input sharpening is added at default but the user needs to activate things like noise reduction and preferably come up with one's own presets for those. Like in Capture One, you can create presets for each of the modules and come up with your own processing styles which can be applied to the entire batch of images on import.

Click to expand...

So, are you saying that if you activate it, Darktable will provide optimised noise reduction and sharpening for a specific camera and ISO, as C1 does by default (optimised in terms of what Phase 1 think it means of course)

Click to expand...

The darktable NR profiling is based on the (quality of) sample images supplied by the open source community, so I imagine the results can't be as consistent as is the case with a commercial support. But yes, NR optimization is at your disposal when your camera is supported in dt.

sankos said:

The darktable NR profiling is based on the (quality of) sample images supplied by the open source community, so I imagine the results can't be as consistent as is the case with a commercial support. But yes, NR optimization is at your disposal when your camera is supported in dt.

Click to expand...

It will be interesting to see how the OP gets on, when he does the comparison again with optimisation switched on.

Though I do find that what different people consider to the optimal compromise between noise and sharpness differs tremendously.

Cheers,

Graham

sankos said:

Your darktable conversion contains chroma noise which you haven't addressed. Create another instance of the profiled denoise module using the wavelets mode and the color blend(ing) mode. See this post for screenshots.

Also, consider applying (large radius) sharpening only to those areas that really need it, so use masking in your sharpening module (or don't sharpen at all).

Click to expand...

Using wavelets in the color blend denoise was a great hint, thank You so much! Also, desaturating the red channel made the remaining color noise much less prominent. So I already learned something

myotisone said:

It will be interesting to see how the OP gets on, when he does the comparison again with optimisation switched on.

Though I do find that what different people consider to the optimal compromise between noise and sharpness differs tremendously.

Cheers,

Graham

Click to expand...

The profiled noise reduction was on, but I did not know how to best address the color noise specifically.

But I am happy that my suspicion was right and this was a PEBKAC. I like that Darktable is always telling what it does, and prefer learning new stuff to handholding.

Speaking of profiles, maybe I should read up on how to contribute samples (or amybe even calibration) of the SEL18135 to lensfun. That its distortion correction was not yet available was the original reason I installed C1.

A J Fresnel said:

The profiled noise reduction was on, but I did not know how to best address the color noise specifically.

But I am happy that my suspicion was right and this was a PEBKAC. I like that Darktable is always telling what it does, and prefer learning new stuff to handholding.

Speaking of profiles, maybe I should read up on how to contribute samples (or amybe even calibration) of the SEL18135 to lensfun. That its distortion correction was not yet available was the original reason I installed C1.

Click to expand...

I don't know much about Darktable, only read good things about it, or lensfun.

The C1 approach is to get you to a "good" result quickly, and use that as a basis for further editing. I suspect its related to it's origins as a program designed for tethering to Phase One cameras, where the client is potentially looking at the pictures as they pop up on the computer.

It's still just a starting point, and you can build your own noise presets and edit the C1 results to taste. As I said in my post to Sankos, everyone has a different idea of where the best balance of noise to detail lies in terms of noise reduction. I have found the defaults of all specialist noise reduction programs I've used (e.g. DXO prime, Neat Image etc) to be horrible.

But I'm a great fan of Open Source software, even though I don't use any photographic examples of it, and glad you have sorted out Darktable.

Cheers,

Graham

myotisone said:

A J Fresnel said:

The profiled noise reduction was on, but I did not know how to best address the color noise specifically.

But I am happy that my suspicion was right and this was a PEBKAC. I like that Darktable is always telling what it does, and prefer learning new stuff to handholding.

Speaking of profiles, maybe I should read up on how to contribute samples (or amybe even calibration) of the SEL18135 to lensfun. That its distortion correction was not yet available was the original reason I installed C1.

Click to expand...

I don't know much about Darktable, only read good things about it, or lensfun.

The C1 approach is to get you to a "good" result quickly, and use that as a basis for further editing. I suspect its related to it's origins as a program designed for tethering to Phase One cameras, where the client is potentially looking at the pictures as they pop up on the computer.

It's still just a starting point, and you can build your own noise presets and edit the C1 results to taste. As I said in my post to Sankos, everyone has a different idea of where the best balance of noise to detail lies in terms of noise reduction. I have found the defaults of all specialist noise reduction programs I've used (e.g. DXO prime, Neat Image etc) to be horrible.

But I'm a great fan of Open Source software, even though I don't use any photographic examples of it, and glad you have sorted out Darktable.

Cheers,

Graham

Click to expand...

i've been working a while now with Darktable, and it enables great workflow, as soon as you create your own presets. I compared it with C1 for both RAW and JPG. Since I shoot nature, and preferably action, with variable lighting conditions, and non-ideal conditions, where you cannot do a "retake", I usually prefer JPG to RAW (did not see the big difference in the output JPG.

However, in Darktable, I found that the input profile has a profound influence on the results, and also the "base curve". In Darktable you get a single generic 'base curve' for Sony cameras - whatever the device, while for C1 e you get the customized one for the model you use. I actually tried my A9 preset from C1 on an A7Riv photo, and it made a distinct difference in sharpness and noise. The same cannot be said for Darktable at this time. Also, in both JPG and RAW, the C1e output seems to yield a more popping result (in that for instance, the birds eye was more pronounced and looked fetching, while the feather details where similar between the two.

Having "skin in the game", as it were, when referring to post-processing SW, is I feel somewhat crucial both in understanding the question behind the question, and in providing ways to reach the desired end result, or just saying - "Hey, I tried, but I don't really see how to help you" - and that's a fine an answer as can be.

Can't really tell, because the image itself is unsharp and has motion blur.

CO does desaturate it a bit, and the chroma noise is less obvious, but it is still there.

Do you have the raw file for us to play with, so we could have a go and see where we get at?

SimpleUser said:

i've been working a while now with Darktable, and it enables great workflow, as soon as you create your own presets. I compared it with C1 for both RAW and JPG. Since I shoot nature, and preferably action, with variable lighting conditions, and non-ideal conditions, where you cannot do a "retake", I usually prefer JPG to RAW (did not see the big difference in the output JPG.

However, in Darktable, I found that the input profile has a profound influence on the results, and also the "base curve". In Darktable you get a single generic 'base curve' for Sony cameras - whatever the device, while for C1 e you get the customized one for the model you use. I actually tried my A9 preset from C1 on an A7Riv photo, and it made a distinct difference in sharpness and noise. The same cannot be said for Darktable at this time. Also, in both JPG and RAW, the C1e output seems to yield a more popping result (in that for instance, the birds eye was more pronounced and looked fetching, while the feather details where similar between the two.

Having "skin in the game", as it were, when referring to post-processing SW, is I feel somewhat crucial both in understanding the question behind the question, and in providing ways to reach the desired end result, or just saying - "Hey, I tried, but I don't really see how to help you" - and that's a fine an answer as can be.

Click to expand...

With this thread started in 2018, I'm just wondering if you're playing with the latest version of darktable, as its development has seen rapid changes in the past year. They're shifting to scene-referred workflow, which be more powerful and simpler (since you mentioned base curve, and should be using filmic instead).

https://darktable-org.github.io/dtdocs/overview/workflow/edit-scene-referred/

SimpleUser said:

i've been working a while now with Darktable, and it enables great workflow, as soon as you create your own presets. I compared it with C1 for both RAW and JPG. Since I shoot nature, and preferably action, with variable lighting conditions, and non-ideal conditions, where you cannot do a "retake", I usually prefer JPG to RAW (did not see the big difference in the output JPG.

Click to expand...

Strangely, I downloaded Darktable this morning to give a try.

My fist impressions, unfortunately are that its unusably slow on my 2011 Mac Mini. It wasn't possible to scroll through the 117 files I imported as they just jumped in block up and down the screen rather than scrolling.

In the end I gave up, but will give it another go tomorrow.

Graham

myotisone said:

Strangely, I downloaded Darktable this morning to give a try.

My fist impressions, unfortunately are that its unusably slow on my 2011 Mac Mini. It wasn't possible to scroll through the 117 files I imported as they just jumped in block up and down the screen rather than scrolling.

In the end I gave up, but will give it another go tomorrow.

Click to expand...

It might be that upon importing, darktable has to parse and generate thumbnails for all of them, so it'll be unresponsive the first few seconds or minutes while doing so. Depending on your hardware (memory, HDD speed), this might be long. A decent GPU will help.

Perhaps a 2011 Mac Mini is indeed too old, especially if you keep upgrading your cameras to more megapixels.

I'm still on my cheap 3-yo Dell laptop, i5-7300HQ with 16GB RAM, 256GB SSD, entry-level GTX 1050Ti, and darktable is buttery smooth handling 24MP RAF files. The GTX 1050Ti helps a lot though, speeding darktable up 5-8 times comparing to the i5 alone.

sluggy_warrior said:

myotisone said:

Strangely, I downloaded Darktable this morning to give a try.

My fist impressions, unfortunately are that its unusably slow on my 2011 Mac Mini. It wasn't possible to scroll through the 117 files I imported as they just jumped in block up and down the screen rather than scrolling.

In the end I gave up, but will give it another go tomorrow.

Click to expand...

It might be that upon importing, darktable has to parse and generate thumbnails for all of them, so it'll be unresponsive the first few seconds or minutes while doing so. Depending on your hardware (memory, HDD speed), this might be long. A decent GPU will help.

Perhaps a 2011 Mac Mini is indeed too old, especially if you keep upgrading your cameras to more megapixels.

I'm still on my cheap 3-yo Dell laptop, i5-7300HQ with 16GB RAM, 256GB SSD, entry-level GTX 1050Ti, and darktable is buttery smooth handling 24MP RAF files. The GTX 1050Ti helps a lot though, speeding darktable up 5-8 times comparing to the i5 alone.

Click to expand...

I did give it what I thought was plenty of time to build the previews, but my Mac Mini is indeed a bit old (slow to render previews with LR, but instantaneous with C1, which is one of the reasons I am still with C1).

The files were 24mp NEFs and 16mp RAFs

The main problem with the mini as its actually reasonably fast, i7 2.6? gHz with an SSD and 16Mb of RAM. Its the GPU that is the issue with only 256MB of RAM.

I will need to upgrade soon, but as C1 and PS run fine I keep on putting it off.

Having said that, I have reached the end of the road with PS upgrades as my Mac is too old to run an OS that supports the latest PS release.

Graham

myotisone said:

I did give it what I thought was plenty of time to build the previews, but my Mac Mini is indeed a bit old (slow to render previews with LR, but instantaneous with C1, which is one of the reasons I am still with C1).

The files were 24mp NEFs and 16mp RAFs

The main problem with the mini as its actually reasonably fast, i7 2.6? gHz with an SSD and 16Mb of RAM. Its the GPU that is the issue with only 256MB of RAM.

Click to expand...

hmm, the i7 2.6GHz should be plenty fast for darktable, the GPU only speeds up operations you apply to an image (denoise, demosaic, ...). I assume that you're starting fresh with no previous darktable cache/config?

Another possibility is that if the folder has sidecar XMP files from other software (e.g. LR), darktable may try to import them, and thus, slow down. I guess you can just try importing from an empty folder with only 5-10 NEFs or RAFs and see if that will be different.

If you still have patient, you can also run "darktable -d perf" from a terminal, it'll print out performance details to help with pin-pointing the problem.

sluggy_warrior said:

hmm, the i7 2.6GHz should be plenty fast for darktable, the GPU only speeds up operations you apply to an image (denoise, demosaic, ...). I assume that you're starting fresh with no previous darktable cache/config?

Another possibility is that if the folder has sidecar XMP files from other software (e.g. LR), darktable may try to import them, and thus, slow down. I guess you can just try importing from an empty folder with only 5-10 NEFs or RAFs and see if that will be different.

If you still have patient, you can also run "darktable -d perf" from a terminal, it'll print out performance details to help with pin-pointing the problem.

Click to expand...

All my raws have XMP sidecar files (created by Photo Mechanic), but I'm not finding the import slow, it's the scrolling through the thumbnails that are impossible to navigate.

Having said that, I've just tried again, and things work fine with the arrow keys, it's the mouse that is giving problems.

Graham

sluggy_warrior said:

SimpleUser said:

i've been working a while now with Darktable, and it enables great workflow, as soon as you create your own presets. I compared it with C1 for both RAW and JPG. Since I shoot nature, and preferably action, with variable lighting conditions, and non-ideal conditions, where you cannot do a "retake", I usually prefer JPG to RAW (did not see the big difference in the output JPG.

However, in Darktable, I found that the input profile has a profound influence on the results, and also the "base curve". In Darktable you get a single generic 'base curve' for Sony cameras - whatever the device, while for C1 e you get the customized one for the model you use. I actually tried my A9 preset from C1 on an A7Riv photo, and it made a distinct difference in sharpness and noise. The same cannot be said for Darktable at this time. Also, in both JPG and RAW, the C1e output seems to yield a more popping result (in that for instance, the birds eye was more pronounced and looked fetching, while the feather details where similar between the two.

Having "skin in the game", as it were, when referring to post-processing SW, is I feel somewhat crucial both in understanding the question behind the question, and in providing ways to reach the desired end result, or just saying - "Hey, I tried, but I don't really see how to help you" - and that's a fine an answer as can be.

Click to expand...

With this thread started in 2018, I'm just wondering if you're playing with the latest version of darktable, as its development has seen rapid changes in the past year. They're shifting to scene-referred workflow, which be more powerful and simpler (since you mentioned base curve, and should be using filmic instead).

https://darktable-org.github.io/dtdocs/overview/workflow/edit-scene-referred/

Click to expand...

Dear SW,

I found Darktable last year, in working with GIMP. After some deliberation, I saw that it could do for me all that I wanted from GIMP, and do it better and faster, and its workflow was better.

I'm using ver. 3.4, so you're welcome to recommend any changes to what I have been working with.

However, when switching to Darktable as primary for both RAW and JPG, I was introduced to Capture One Express. Although it is limited in some annoying ways (lack of watermark/ retouch for instance..), it has some great abilities that Darktable doesn't. Even my ability to crop freehand and then correct the measurements on the sidebar. Very easy and intuitive, and to date, after switching fully to Darktable, when trying it out on some new RAW files and comparing to Capture One Express, I still get better results from the Capture One. Don't know exactly what they did, but I remember a half a year ago, when trying the same (shooting RAW & JPG and comparing DT to C1e) that I came to the same conclusion. C1e did something very right, to produce easily popping images.

-Still debating - Darktable is a great tool.

SimpleUser said:

However, when switching to Darktable as primary for both RAW and JPG, I was introduced to Capture One Express. Although it is limited in some annoying ways (lack of watermark/ retouch for instance..), it has some great abilities that Darktable doesn't. Even my ability to crop freehand and then correct the measurements on the sidebar. Very easy and intuitive, and to date, after switching fully to Darktable, when trying it out on some new RAW files and comparing to Capture One Express, I still get better results from the Capture One. Don't know exactly what they did, but I remember a half a year ago, when trying the same (shooting RAW & JPG and comparing DT to C1e) that I came to the same conclusion. C1e did something very right, to produce easily popping images.

Click to expand...

I only use free/open-source software so I can't compare it to C1. Maybe share a RAW and its JPG exported from C1, I'll try to match the output using darktable. If I couldn't with my limited experience of darktable, then perhaps darktable is indeed still not very easy to use.