I agree for the destructiveness of using layers, but I don't feel it's that much of a problem. Also doing that, I personally would use the "original" model sheet as my base image en retransform it instead of transforming an already transformed frame. Or you just reimport the ref. But I feel it's much more secure and efficient to place an image at each position you like and keep it like that instead of messing with te relative positions of the lighttable, you always get the right out of peg at the right time.
I also don't have pb with messy layer opacity with the "see only current layer" option. I desactivate the layers I dont want to see and keep the one I use as perspective guides (or all except current isteand of dimming). You do it once and then you keep on animating, it's no big deal to me. Honestly we don't have lots of layers in our projects anyways (I usually use a 3-5% setting, 7% max, on the layer opacity setting of show only and I'm fine with it. Same opacity I use with LT). I also like to desactivate my animation layer so when I flip up down, it dissapears completely and rappears when I go back to drawing
For non-destructive workflows, I use guidelines, placing multiple guides and using my guideline toggle switch script to toggle between the positions. It has the non destructive benefit + it's not on a new layer and ca go out of the canvas (you just need to switch guidelines regularly until you find the one you like). Here are the scripts :
https://forum.tvpaint.com/viewtopic.php?t=15615
You can also use the brush to image guide to reference another layer you like and add it as a nondestructive image in your image guides.
If you really like using the lighttable, you can also put your guide as an instance at the end of your current layer (after out point for ex), put lighttable mode in "bookmark" or "imagemark" mode, and mark your ref. Then you use only the +1 OOP of your lighttable to manipulate that faraway image on each image, with the LT image not switching position when you flip from one drawing to another. With this method with LT you can also put multiple marks at the end for your ref image, and use them as different out of pegs, using the different positions of the lighttable to switch between ref +1 to ref+10. I also don't want to go back to my scripts again, but I feel I'd like to use this method because I can use my OOP scripts to easily position the LT drawing
The difference maybe with this one is that you don't get the benefit of the first method with layers I talked about in my first paragraph = you can place each out of peg in time and retain the position on each image cause it's an actual instance synced with your animation.
I don't really see an obvious reason for this layer lightable mode that would be really easier than the current options, but maybe a separate interface with layer references/opacity states/non-destructive repositioning of layers would be options that could benefit workflows even for storyboard or something. If you think about it, layer opacities are already kind of a vertical lightable and the only thing we miss in non-destructively being able to reposition layers in time like keyframing in AE
Another fun option we could ask the team is to add an option to the fx stack to apply it on layers of only certain color groups. The we would use the fx stack keyframer with preview on and "apply on layers of this color only" to kind of out of peg it non destructively
I tried with "apply on current group" but of course you can't really make it work that way :')
)
