Doom Action Prompter

Naming again, because it matters before you start searching: the class behind this page is DoomActionPrompter, and the node shows up in your search bar as Doom Action Prompter. KontextPrompterUltimate_RU is the legacy alias kept for old workflows. You place Doom Action Prompter; you're reading about it because the alias still gets searched.

What it is: a dropdown of twenty image-to-image "action" templates. Restyle this, change the background, replace the objects, relight the scene, re-compose the shot, add text - each one is a pre-written prompt template with slots for your specifics. You pick a template, fill in one or two {input1}/{input2} placeholders, and the node hands you a finished prompt-command ready for an I2I sampler or an instruction-editing model like Kontext.

How it works

• action_choice - the dropdown, formatted Category >> action. The default gives you the shape: "Композиция и Кадрирование >> Изменить ракурс на: [например, 'вид сверху', 'снято снизу']" (composition and framing → change the angle to...). Each template is a Russian sentence with placeholders where your specifics go.
• input_1 / input_2 - the substitution values for {input1} and {input2} inside the chosen template. Not every template uses both; the tooltips tell you. The default template uses only {input1} (the new angle).
• prompt (output) - the assembled, ready-to-use command.

The workflow is: pick the action you want to perform on your image, type the subject into the placeholders, get a prompt that reads like an instruction. For Russian-speaking users this is the fastest way to translate "I want it to look like a 90s anime still" into a structured prompt-command without drafting the sentence from scratch.

Setup

Standard pack install:

cd ComfyUI/custom_nodes
git clone https://github.com/PeterMikhai/Doom_Flux_NodePack

Restart ComfyUI, or search "DoomAI Nodes" in Manager. No dependencies.

Where people get burned

• The alias is not a separate node. Both names resolve to the same Doom Action Prompter. If a saved workflow references the _RU name, it still loads - but searching for the current display name is the way to place it fresh.
• The output is Russian and it's an instruction. It's built for the pack's Russian-speaking audience and for models that read instructions (Flux, Kontext-style editors). If your model wants English, run the output through Doom Translate first.
• Templates have fixed slots. {input1}/{input2} are the only blanks. A template that doesn't match your need can't be edited mid-node - you'd write your own prompt instead. The twenty templates cover the common I2I moves, not every conceivable one.
• It's a prompt, not a pipeline. This node only produces the text. The actual img2img/inpaint/edit still happens in your sampler - wire the prompt output into whatever model does the work.

It's a niche tool - a Russian-language, image-to-image prompt-command library in a dropdown - and it knows exactly what it is. If you're its audience, it saves you from drafting the same five I2I sentences for the hundredth time.