I have just begun using Darktable 4.0. I have set to modern, filmic pipeline, and left the default modules untouched (input & output color profile module defaults). I am editing and applying a custom style to a series of images that will be used as ‘pseudo’ stopmotion & combined with footage in Cinelerra.
So far I am able to export in jpeg 8 bits, using the project’s colour settings (output color profile module’s default is sRGB). If I try to change the jpeg output profile to rec709, there is posterisation wether in gwenview or the Gimp (converted to sRGB). I haven’t had a chance to test in Cinelerra but the posterization seems odd.
I thought it might be useful to also export TIFF, as I might need to chromakey out the background for some images, so I figure that the TIFF format would allow for better manipulation? I’m not sure if I should also save to rec. 2020 and let the video editor reduce to rec. 709 upon rendering?
Can anyone recommend the best colourspace workflow for input profile, darktable workspace (keep at default rec.2020?) and export settings specifically when wanting to edit along with video footage in a color managed video editor. I assume it needs to be rec. 709?
My images are larger than needed so I try to export to (1920 X1080) pixels but they export to 1618 X 1080. I assume their is a keyboard shortcut to force the dimensions or ratio?
darktable export doesn’t crop by default. It’s scales the image so it fits to the limiting parameter (hight or width). If you need full hd export you need to crop the image to 16:9 yourself.
are you sure the used viewers are doing the conversion properly? If you need rec709 stuff it doesn’t make sense to check it in an application that isn’t able to handle it. Better import the exported image in darktable since darktable is able to properly respect the embedded profile…
I would set your export and output profile to whatever your working profile is in the video editor. If it’s rec 709, use that. If rec 2020, use that.
If you export jpeg you are compressing the image. If you export tiff you are not. Therefore jpeg is meant as a display format (ie. Finished product) while tiff is meant as a working format (ie. More work to be done). So if you are doing Chroma key to remove green screen, tiff is preferable. Also, I gather you will be compressing the video when done. If so, you will just be adding compression on top of compression for the jpeg images, which is not ideal.
However, jpeg will have much smaller file sizes and could be faster to work with, which might be what you need, pending your hardware.
Png is an alternate to tiff if you need smaller file sizes. They can also be saved without compression.
Depending on what you are trying to do in DT…you could just convert them to DNG and load them strait into Davinci Resolve. you can do some pretty nice edits on them and then DR will handle all the colorspace and resizing business and you won’t have to go through all the work in DT and the exporting but I have no idea maybe you need some aspect of DT for the images…
Thanks everyone for the info, I always wanted to work with RAW, so this project is a good introduction.
I found out from Aurélien Pierre, darktable dev., that 8 bit jpeg is not encoded to avoid posterization with rec 709, best to use sRGB. Since I do want to have the option of furthur editing, I will keep it in Tiff 16 bits, since Cinelerra can create intermediate codecs if needed, but I will keep the PNG as an option.
I think for this project I will stick with Cinelerra GG, because I am more familiar with it, but I would like to experiment with Davinci Resolve and Lightworks.
If I understand it properly; Cinelerra workspace offers up to RGBA float (32 bits) or YUVA 8 bits, but it displays in sRGB. So its best to use RGBA float for the project, with the YUV color space choice of Rec709 & YUV color range to JPEG.
Thus my pictures should be exported to Tiff 16 bit Rec709?
I suggested Davinci because you can actually get pretty nice edits with stills. They have a color match feature which works amazingly and you can basically transfer the look of any image to a whole series…and it can manage all the colorspace changes back and forth with a simple node so for me I would not mess with intermediate files in another program unless there was some fairly serious advantage to doing an edit there and then transferring that information. A dng would be as good as any tiff and Davinci even has a couple of auto tools that can produce some nice results… there may be some good reasons to do what you are doing but if the destination is video you might at least consider trying to do it all in Davinci…