Is there a way to expand the canvas to the right by 300%?
As far as I understand the answer in the thread linked to, the answer is yes.
(Havenāt ver. 4.7 myself, so I cannot confirm.)
Iām running 4.7.0+710~gb7511e7f56 and I can only expand 100%. Maybe there is some way unbeknown to me on how to do it.
If thatās the case, maybe what @arturoisilvia did was to expand 100 percent and add image no. 2. Then saving the composite and then expand 50 percent and add image no. 3?
Donāt understand, sorry!
Arturosilvia posted a triptych in the post I linked to above. This was presented as a response to my question whether we can use the new module to create diptychs and triptychs.
So if itās true that one can expand only 100 percent, as you say, Iām wondering if he perhaps first doubled the canvas and added a second image, exported the composite image, and then opened the new file, expanded it so as to add the third image and frames?
1 Like
Aha, yes that work. Thanks!
And there is also the aspect what can be achieved from that ācontent can be scaledā?, ref the draft Release notes: darktable/RELEASE_NOTES.md at master Ā· darktable-org/darktable Ā· GitHub
If one can expand canvas to the double, and then scale down images applied to it, this may perhaps also be a way to achieve triptychs. You who have v. 4.7 could perhaps look into this.
1 Like
While doing such things as diptychs and triptychs within darktable is nice, I have to wonder if it isnāt going a bit beyond whats needed, the more so as there are programs that allow this kind of operations already (Gimp, Krita, probably others). And with them, you have many more possibilities (e.g. with respect to framing etc.).
As this thread is about the ācanvas enlargementā module, do you really want the expansion to be independent of the rest of the image? Or is that a sign you are abusing the module beyond its design limits? (In hte specific case of diptychs and tryptichs, youāre probably also pushing the overlay module beyond itslimitsā¦)
Darktable is a program with a linear pixel pipe. As such, anything done within that pipe will be influenced by everything later in the pipe, unless you start using masks.
The program @Pieter_Zanen quoted is based on a node graph wich allows parallel paths. Thatās a rather different implementation, with a different basic design behind it. While independent processing of two parts of an image is fairly natural for such a system, forcing that concept into darktable wonāt be pretty, if it even is doable.
Indeed, that is how I did it.
As long as you like the result, why not?
See, another case of abuse? Or do you like it? 
:
May be.
Iāve just wanted to know what the new modules can be used for, as creating diptychs/triptychs are the only thing I currently feel the need to access Gimp or any other image program outside dt for.
Thanks for the input to such clarification!
EDIT: Threr is already a sneaky method to make overlays by āabusingā the watermark module. As long as it works, there seems to have been those who have been happy to use it for e.g. changing skies
In this particular context, the issue is not so much the āabuseā of the module in itself, but the remark about not wanting the added area to be influenced by subsequent modules in this particular use. But next time, you may want to add a bit of background at one side and then you do want subsequentmodules to act on the added areaā¦
In essence, the exemple (di- and triptychs) would require complicating the module for one particular use case. There are already very suitable tools available for that use case, and they offer many possibilities to do this.
Well, Iāll keep all that in mindā¦ thanks.
I donāt immediately see any need for complicating the module to employ it for creating diptychs. From what Iāve learned so far, we can use it as it is, (although the modules would be more useful if the expansion limit of 100 percent was expanded somewhat).
To use any images developed in dt in a composite image created with another program we need to export them as .jpeg, .tiff or whatever, and then import the images into a compositing program where the images are put onto a common canvas and finally exported as one composite image file.
With the new modules available (really looking forward to v. 4.8 
), I foresee a similar workflow but within dt only:
Develop the first image, create a snapshot of the result, and export. Develop the second image ā using the snapshot of the first image for guidance in scaling/positioning/grading ā and export it. Add the two exported image files to dt library and open the first in darkroom. Expand the canvas and then overlay the second image file besides the first, add a frame by the Framing module, (if we cannot expand the canvas for this purpose also), and export the finished composite.
In such a use case there is no problem that the canvas color will be influenced by any subsequent editing, because we will do no kind of grading after the overlay.
If it is not possible to expand the canvas beyond 100 percent, but one will anyhow want some frame/spacing between the two images, this could be taken care of by creating a mask with a rectangular field positioned on the border between the two images with the Retouch module, and fill it with the color of the frame around.
This seems fairly straight forward to me, but you mentioned something about āprobably also pushing the overlay module beyond its limitsā that Iām not aware of what is referring to.
EDIT: If one also make a snapshot of the first edited image in a reverse orientation, it will be even easier to see where lines in the two images meet.
That workflow may not do what you want for any space between the elements of the composite.
the benefit of this module is to be able to apply the same color / sharpness /ā¦ edits to a composite image. And this is also the weak point over pure image editing software: the same edits are applied to the whole composite if not limited manually by masks.
If youāre aware of this, you can make use of this in darktable - if not youād better do this in a dedicated image editing tool like krita or gimp
In fact, using ChaiNNer it is quite easy to merge several photoās including borders. See the enclosed scrrenshot
Just a few remarks about ChaiNNer from what I read in their docs:
- itās still in alpha stage, so no guarantees about stability and compatibility.
- it uses floats for image representation, with a range clamped to [0ā¦1], a bit like working in display-referred in darktable
- Itās not actually an image editor: quote from chaiNNer/docs/nodes.md at main Ā· chaiNNer-org/chaiNNer Ā· GitHub
ChaiNNer is a visual programming language and development environment.
It could be a very interesting program, but presenting it as a viable alternative for existing programs feels a bit prematureā¦