Nodes/arkennemasis - ChatGPT / Codex / fal AI nodes/arkennemasis fal · Flux Pro Kontext Text to Image · $0.04/image
ComfyUI Node

arkennemasis fal · Flux Pro Kontext Text to Image · $0.04/image

Kontext's prompt-following without the reference photo

By Hishamahmer·Created 3 months ago·Updated 9 days ago· 11
arkennemasis fal · Flux Pro Kontext Text to Image · $0.04/image
    • images
    • image_paths
    • seed
    • has_nsfw_concepts
    • prompt
    • timings
    • info
    ◄prompt►
    ◄seed-1►
    ◄guidance_scale3.50►
    ◄num_images1►
    ◄output_formatjpeg►
    ◄safety_tolerance2►
    ◄enhance_promptfalse►
    ◄aspect_ratio1:1►
    ◄max_cost_usd20.0►
    ◄reuse_identical_runtrue►
    ◄extra_json►
    ◄max_concurrent1►

    Straight text-to-image on the Kontext Pro endpoint, $0.04 per image. Worth knowing what this actually is before you wire it up: Kontext's reputation is editing, and the interesting thing about the text-to-image variant is that it inherits the same prompt-following and in-image text rendering - just without a reference to edit. Think of it as the mid-tier alternative to local Flux when you can't run 12B or don't want to.

    The inputs

    • prompt - write it as a description, in sentences.
    • aspect_ratio - defaults to 1:1. Options: 21:9, 16:9, 4:3, 3:2, 1:1, 2:3, 3:4, 9:16, 9:21. Pick deliberately; a landscape scene in a square frame is a choice you probably didn't mean to make.
    • guidance_scale 3.5 (1–20) - adherence to the prompt. seed for repeats, num_images 1–4.
    • output_format (jpeg/png), enhance_prompt (off by default; fal rewrites your prompt when it's on), safety_tolerance default 2 with 1–6 available, 1 strictest.

    And the pack-wide widgets that come on every fal node here: max_cost_usd (default $20 - checked before upload, so a bad setting can't bill you), reuse_identical_run (on; identical settings plus identical connected media return the earlier files), extra_json (JSON merged last into the request, for anything the node has no box for), max_concurrent (default 1 - one paid fal call at a time across the run).

    Outputs

    images → preview or save. image_paths → the files under output/fal/fal-ai_flux-pro_kontext_text-to-image/. seed for reproducibility, prompt echoing the text sent, has_nsfw_concepts, timings, and info with the request id, cost estimate and fal's answer.

    Install

    cd ComfyUI/custom_nodes
    git clone https://github.com/Hishamahmer/comfyui-arkennemasis
    

    Or search comfyui-arkennemasis in ComfyUI-Manager. Restart, then find it at arkennemasis/fal/Image/Flux. Nothing to download locally.

    # .env, beside run_nvidia_gpu.bat
    FAL_KEY=your-key-from-fal.ai/dashboard/keys
    

    No key field on the node - the pack reads the .env file and nothing else, so an FAL_KEY set as an environment variable won't be picked up.

    Worth knowing before you spend

    Pro is closed weights. There's no local download at any VRAM, which is the honest justification for paying per image here at all. If you can run Flux Dev locally, the Pro route is mostly a convenience and a quality-tier choice rather than a necessity.

    No acceleration or step controls. This endpoint gives you prompt, guidance, ratio, seed and count. Iteration means rewriting the prompt, not turning a dial - which is a real difference from the local Flux experience where you'd sit there adjusting steps.

    Text in images is a strength here, and it's the main reason to reach for Kontext's t2i over other hosted endpoints: signs, labels and short headlines come out legible far more often. Long strings still break.

    Censorship lives in the model, not the node. The Kontext-era license and the weights put explicit content off the table; raising safety_tolerance doesn't change what the model will render.

    Categoryarkennemasis/fal/Image/Flux

    Inputs (12)

    NameTypeDefaultDescription
    promptSTRINGThe prompt to generate an image from.
    seedINT-1-1–2147483647The same seed and the same prompt given to the same version of the model will output the same image every time. -1 leaves it to the model.
    guidance_scaleFLOAT3.501–20The CFG (Classifier Free Guidance) scale is a measure of how close you want the model to stick to your prompt when looking for a related image to show you.
    num_imagesINT11–4The number of images to generate.
    output_formatCOMBOjpegThe format of the generated image.
    safety_toleranceCOMBO2The safety tolerance level for the generated image. 1 being the most strict and 5 being the most permissive.
    enhance_promptBOOLEANfalseWhether to enhance the prompt for better results.
    aspect_ratioCOMBO1:1The aspect ratio of the generated image.
    max_cost_usdoptFLOAT20.00–100000Safety cap. If the estimated cost of this run is above this many US dollars the node stops BEFORE sending anything to fal. 0 = no cap.
    reuse_identical_runoptBOOLEANtrueIf these exact inputs (the settings AND the same pictures/clips/audio) were already paid for and the files are still in output/fal, return them instead of paying again - also after a ComfyUI restart. Turn off for a fresh take with the same settings.
    extra_jsonoptSTRINGAdvanced: a JSON object merged into the request last, for any fal field this node has no box for. Example: {"seed": 7}. Leave empty normally.
    max_concurrentoptINT11–32How many paid fal calls may run at the same time in one run, across every fal node. 1 (default) = one after another. This node starts its paid part only while fewer than this many fal calls are running. Once any fal node fails, the ones still waiting are not started, so nothing more is billed.

    Outputs (7)

    NameTypeDescription
    imagesIMAGE—
    image_pathsSTRING—
    seedINT—
    has_nsfw_conceptsSTRING—
    promptSTRING—
    timingsSTRING—
    infoSTRINGJSON: request id, cost estimate, saved files and fal's full answer.