I have older photos that I shot in JPEG-only. I know DT can edit them, but what color spaces are available?
As I understand, opening a JPEG, the modules work in a display-referred color space. So you can cause white / black clipping in one module that you’ll never un-do in a higher module.
If you activate filmic RGB or sigmoid, do the modules below them work in the scene-referred color space for a JPEG?
An easy example is to turn on the clipping indicator and add exposure to a JPEG to push the whites over-exposed. When I activate filmic RGB, the white clipping false-color disappears.
The rationale seems like it works. Is that what’s actually happening?
THe jpgs are in a sense already clipped/mapped and constrained . So there is no real benefit of scene-referred. You aren’t able to create new data by pushing them beyond that and have scene referred modules that will magically recover. On the other hand if the raw file presents linear light clipped only by the limits of the sensor then there are benefits to be gained to preserve as much of that data as is possible. THis is a big oversimplification but IMO you can’t unwind a jpg to make the scene-referred modules workflow provide any addtional advantage.
I was really asking about not losing data during the edit. Flipping-the-script, if all you have is the JPEG, how do you save everything that you have? There isn’t a rule about being unable to edit JPEG using a scene-referred work, is there?
As I understand display-referred, it is a fixed dynamic range. A module can cause the image to have clipping. Higher-positioned modules can’t bring back the lost data caused by a lower-positioned module. That’s the main down-side about using a display-referred workflow, isn’t it?
As I understand scene-referred, it has “infinite” dynamic range. “Clipping” depends on the settings of filmic RGB or sigmoid. Below these modules, you aren’t going to have data loss caused by a scene-referred module.
This experiment suggests that on a JPEG source, if filmic rgb is off, exposure operates in display-referred, but if filmic rgb is on, exposure operates in scene-referred.
Is that what’s happening? If all you have is a JPEG, are there any down-sides to enabling flimic rgb (or sigmoid)?
Look in the bottom right corner of the darkroom and you will see options for non-raw files. I am sorry I can not do screen shot because I haven’t worked out how to in Linux mint yet.
No rule just that what you are describing at least to me would be that you are pushing a file that is likely already compressed tonally back out beyond its limits and then recompressing it back. So you can edit a jpg for sure but I don’t think the scene-referred are giving you much if any benefit over std tools when you start with a Jpg… IMO
I don’t think exposure changes at all. In display referred workflow no exposure is added by default and in scene there is a default bump… but now all the tone operators including the basecurve come after it later in the pipeline so it should behave the same… it will give a different result if you drag it after those modules but not based on the workflow.
The module order setting which doesn’t seem to change by default when you use a jpg will help to make exposure work correctly. Note that this is using the ‘v3.0 JPEG’ module order; the important part is having exposure and the other tools after input color profile in the pipeline, so it can undo the sRGB encoding of the embedded profile in the jpg
if you’re in a scene referred workflow, you’re not limited to 0-1 range but can push values further during editing. But since your base material is a jpeg there’s no information that can fill the gaps to for example stretch the range to 0-1.5.
But nonetheless filmic or sigmoid will bring back the whole shebang to the 0-1.
So you’re losing information as well as in the display referred workflow but in a different way.
If you’re used to scene referred workflow, you can use the same tools - because it’s more efficient to use stuff you’re used to - but don’t expect magic.
I’ve used filmic or sigmoid on jpegs, and at least in some cases it works quite well - my approach is to enable filmic/sigmoid, then adjust the parameters (mostly contrast and in filmic black/white relative exposures I think) to get as close I can to a ‘passthrough’ state. Then I can proceed with whatever other adjustments I want, with the luxury of not worrying about clipping etc. (not so much anyway). The point for me of first setting up the tone mapper to a ‘flat’ setting is that often jpegs don’t want any more contrast adding, and default settings will add a substantial amount.
On the other hand, traditional methods like rgb curves can work well on jpgs too - i.e. more Gimp-style editing.