An Image Editor with No Actions

Why non-destructive editing tools are not only more powerful, but EASIER to use, and why non-destructive editing's full payoff doesn't occur until it's total and uncompromising.

I'm going to show that non-destructive editing operators are not only fundamentally better and more powerful, but are also massively EASIER to use. And whether you love or hate 2d node-canvas spaghetti, I'll tell you right now, this is not where we are going.

Destructive tools like Photoshop make the user memorize and use destructive-action construction sequences to get where they are going. Everything is: assemble-products, perform destructive action, if wrong: undo and repeat until correct, flatten into result, move on to next part of the construction. In order to assemble a final output, the user has to mentally work backwards through all the things they will need to do to get there  --- and then start at the beginning! It's like learning to recite the alphabet backwards. How many of you can do that? And Photoshop is many times more complicated than the alphabet. 

However, there is an easier way. My tool has no "actions". There are no configure dialogs where you "Commit" and change the underlying data. I don't need "Photoshop Actions" to replay a sequence, because everything is live through the pipeline always. I have not even added a flatten feature (*yet). And yet most things *feel* mostly layer-like. It's like... better layers.  (* I will eventually add Flatten, but destroying the chain will be your choice, never required to get something done)

Non-destructive operators allow the user to explore and learn. Building something and refining it step by step. One can assemble the final production in any order, because every part of the construction remains live at all times. If you want to mask something, you just create a mask operator "thing", and look at it, and see what it needs, and supply something to its inputs, and see what it produces. The learning becomes EXPERIENTIAL instead of a library research project and memorization.

Below is a quick screenshot of my new easy to use image creation tool.

This sample workflow is taking an input image that's a color "PDF" and converting it to black/white, then piping it into 3 different mask operators (one for each letter) over three different input images, each to fill one of the letter shapes, then merging together the three composed letters into the final image. 

In Photoshop or Affinity, this requires destructive actions and/or duplicating image layers, which breaks the live connection. When you want to change early inputs, you get to repeat the steps. 

In my tool, you can edit any part of the entire image flow, including the original PDF mask image, and everything updates live. You can't do that in Photoshop at all. You can do this in an image compositor, like Nuke, or Mari, or Blender - but then you're working with 2d node-graphs, not something that feels like layers.

My tool is not the first to use non-destructive editing. 

Even among destructive tools, the availability of non-destructive contexts is growing. Photoshop has vector shape layers. Pixelmator has a set of non-destructive tools, and Affinity even claims "most" things can be done non-destructively. However, the billing here is deceptive.

Having a vector shape layer in Photoshop, or a non-destructive tool layer in Affinity, does not actually get you non-destructive composition — because the next tool doesn't take the result of a live pipeline, it takes an image. To compose non-destructively, every operator has to accept live results, not flattened ones. Your whole composition is only as manipulatable as its weakest link. Make one destructive flatten, and edits across the boundary are broken.

My tool uses ONLY non-destructive editing. There is no need to "Flatten" anything, ever. You can just keep building and constructing, keeping everything live, always. 

My tool is also not the first to do that. There is a family of node-based image tools including Nuke, Silhouette, Mari, Blender compositing, even ComfyUI. However, they all arrange operations onto a 2d Node canvas. This is powerful when the workflows get very large. However, it's also somewhere between inconvenient and terrible when one wants their screen real-estate to be focused on the art, not the node spaghetti wiring.

A node-based workflow in ComfyUI:

My image editing tool is trying to free node-based power and ease of use from the spaghetti.

My first screenshot was a fully node-based composition program, and you can see, it looks like a traditional layer stack. Here is a simpler one. This is a Vector shape and a text shape, which is then piped in as a mask layer for a loaded image. 

The blue line represents the mask-image pulling its source from out-of-flow. The bright white underline on the koyote-ramen.jpg image says that this layer is not compositing with the chain below. That's it. Everything is live.

I can select and edit the vector shape layer live, while looking at the base vector shapes, or while looking at the final composition. Everything is saved in the file and editable at any time. No flattening.

My tool has a ways to go before it's going to handle practical everyday workflows, because everyone tends to have their favorite pet features they like to use. However, I'm convinced that when I am able to add the majority of features, it will not only be expectedly more powerful for having a non-destructive node-compositing workflow in a layer-shaped package, but it will also be easier to use because of it.