# AI use cases for HEISS UI (tristmeister/HEISS-UI) Repo: [tristmeister/HEISS-UI](https://github.com/tristmeister/HEISS-UI) | Use cases: 5 | Page: https://dedupedai.com/use-cases/tristmeister-heiss-ui/ HEISS UI is a free, open source local front end that sits on top of a ComfyUI install, giving you a prompt box, gallery and model settings in place of the node graph. This file holds every use case of this repo. Each one starts with a "## " heading and ends with its URL. - Generate images from a prompt without the node graph: https://dedupedai.com/use-cases/tristmeister-heiss-ui/generate-images-without-the-node-graph/ - Run an exported ComfyUI workflow as simple controls: https://dedupedai.com/use-cases/tristmeister-heiss-ui/run-your-own-comfyui-workflow-as-simple-controls/ - Change one part of a picture and keep the rest: https://dedupedai.com/use-cases/tristmeister-heiss-ui/repaint-one-part-of-a-picture/ - Upscale any gallery picture to 2K or 4K: https://dedupedai.com/use-cases/tristmeister-heiss-ui/upscale-a-picture-to-2k-or-4k/ - Run the HEISS UI studio in Docker beside ComfyUI: https://dedupedai.com/use-cases/tristmeister-heiss-ui/run-the-studio-in-docker-next-to-comfyui/ ## How to generate images from a prompt without the node graph with HEISS UI Repo: [tristmeister/HEISS-UI](https://github.com/tristmeister/HEISS-UI) | Difficulty: Beginner | Time: 30 minutes | Basis: From the README Install the HEISS UI launcher and get a prompt-to-picture loop running on top of ComfyUI. **Who it is for:** A hobbyist or small creator with ComfyUI installed who wants to make pictures without wiring nodes by hand. **The problem:** ComfyUI's node graph is powerful but slow to learn for everyday prompt-and-look work. **The outcome:** You write a prompt, pick a model, and see the picture render in a browser studio that runs every job on your own ComfyUI. ### Steps 1. Download the zip for your system from the latest release on the HEISS UI GitHub page. 2. Unpack it and double-click Start HEISS UI (.bat on Windows, .command on macOS, .desktop on Linux), or run `npm start` in the folder. 3. Make sure ComfyUI is running; ComfyUI Desktop connects on its own, and other addresses are set under Settings › Connection. 4. Pick one of the 30 built-in model families and press Get all if it reports missing parts. 5. Write a prompt and watch it render. ### Prompt to paste into your agent ```text Clone or unpack the HEISS-UI repo (github.com/tristmeister/HEISS-UI), start it with `npm start` from its folder, and confirm the studio opens in the browser. Then pick from the model menu, enter the prompt , and report the path of the rendered image in the gallery. Verify that the models list is populated, which means ComfyUI is connected. ``` ### Caveat The studio only lists models and samplers after ComfyUI is connected, so a first run with nothing listed usually means ComfyUI is not running yet. Dual-use notice: only use this on software, sites and systems you own or are authorized to test. Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/ URL: https://dedupedai.com/use-cases/tristmeister-heiss-ui/generate-images-without-the-node-graph/ ## How to run an exported ComfyUI workflow as simple controls with HEISS UI Repo: [tristmeister/HEISS-UI](https://github.com/tristmeister/HEISS-UI) | Difficulty: Intermediate | Time: 45 minutes | Basis: From the README Import a graph you built in ComfyUI and use it from the studio with a clean set of controls. **Who it is for:** A ComfyUI user with a custom workflow who wants teammates or themselves to run it without opening the graph. **The problem:** A custom graph is hard to hand to someone who only needs a prompt box and a result. **The outcome:** Your workflow appears in the model menu under Your workflows, with prompt, size, seed and steps as controls, while your graph runs underneath. ### Steps 1. Build and test the graph in ComfyUI, then export it in API format (the regular workflow file also works). 2. Open the Workflow Gallery from the dock and drop the file on it, or paste the JSON. 3. Check the connections under Change connections, then import. 4. Install any nodes the workflow needs, then find it under Your workflows in the model menu. ### Prompt to paste into your agent ```text Using the HEISS-UI repo (github.com/tristmeister/HEISS-UI) running on a local ComfyUI, import the exported workflow file through the Workflow Gallery and review the flagged connections. Report which inputs were guessed, which nodes run code or go online, and whether the workflow appears under Your workflows. Verify that one test render completes. ``` ### Caveat The workflow only shows up under Your workflows after the nodes it needs are installed, so a missing node pack will hide it until you add that pack. Risk, remote script: The import review warns about nodes that run code, touch files or go online, so read those warnings before running a workflow you did not build yourself. Dual-use notice: only use this on software, sites and systems you own or are authorized to test. Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/ URL: https://dedupedai.com/use-cases/tristmeister-heiss-ui/run-your-own-comfyui-workflow-as-simple-controls/ ## How to change one part of a picture and keep the rest with HEISS UI Repo: [tristmeister/HEISS-UI](https://github.com/tristmeister/HEISS-UI) | Difficulty: Beginner | Time: 30 minutes | Basis: From the README Paint over an area of a reference or start image and describe what it should become. **Who it is for:** A creator with a generated or uploaded picture that is nearly right except for one area. **The problem:** Regenerating a whole picture loses the parts that already worked. **The outcome:** Only the painted part is redrawn, and the rest of the picture stays as it was. ### Steps 1. Add a reference or start image in the prompt bar, or use one from the library. 2. Hover the image and press the brush to open inpainting. 3. Paint over the part that should change, then write what it becomes. 4. Adjust Inpaint strength and Edge softness in the sidebar if the edge looks wrong, then run it. ### Prompt to paste into your agent ```text In the HEISS-UI studio (github.com/tristmeister/HEISS-UI), open the image from the gallery, paint a mask over , and set the prompt to . Report the output image path and confirm that the unpainted area is unchanged. Verify that the selected model is one of the 18 image families that inpaint. ``` ### Caveat Inpainting is in beta, is not available for Ideogram 4, MageFlow, Sana, video or imported workflows, and the model menu marks which families support it. Dual-use notice: only use this on software, sites and systems you own or are authorized to test. Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/ URL: https://dedupedai.com/use-cases/tristmeister-heiss-ui/repaint-one-part-of-a-picture/ ## How to upscale any gallery picture to 2K or 4K with HEISS UI Repo: [tristmeister/HEISS-UI](https://github.com/tristmeister/HEISS-UI) | Difficulty: Intermediate | Time: 45 minutes | Basis: From the README Send an existing image through SeedVR2 from the gallery and keep the upscale next to the original. **Who it is for:** A creator who has a good small image and needs a sharper, larger version for print or sharing. **The problem:** Small generated pictures are often too low in resolution for the use you have in mind. **The outcome:** The upscale is saved next to the original, and a slider lets you compare the two. ### Steps 1. Open a picture in the gallery or viewer and press the upscale arrow. 2. Choose 2K or 4K and choose Fast, Balanced or High effort in Settings › Features › Smart upscale. 3. If the SeedVR2 nodes or model are missing, accept the install offer and wait for the download. 4. Compare the original and the upscale with the slider once the countdown finishes. ### Prompt to paste into your agent ```text In the HEISS-UI studio (github.com/tristmeister/HEISS-UI), upscale the gallery image to 4K using Balanced or High effort. Report the saved upscale path and its pixel size. Verify that the upscale is stored next to the original file and that the SeedVR2 nodes were installed or already present. ``` ### Caveat High effort is the slowest and uses the most graphics memory, so start with Balanced on a machine with limited memory. Dual-use notice: only use this on software, sites and systems you own or are authorized to test. Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/ URL: https://dedupedai.com/use-cases/tristmeister-heiss-ui/upscale-a-picture-to-2k-or-4k/ ## How to run the HEISS UI studio in Docker beside ComfyUI with HEISS UI Repo: [tristmeister/HEISS-UI](https://github.com/tristmeister/HEISS-UI) | Difficulty: Advanced | Time: 60 minutes | Basis: From the README Start the studio in a container that serves the gallery from ComfyUI's output folder. **Who it is for:** A developer or home lab user who already runs ComfyUI and wants the studio packaged in a container. **The problem:** Installing and keeping the studio running by hand on a server is fiddly. **The outcome:** The studio runs in a container with its data on a volume, and the gallery shows ComfyUI's output folder. ### Steps 1. Make sure your ComfyUI is running on the same machine. 2. Set COMFY_OUTPUT_HOST to ComfyUI's output folder before starting. 3. Run `docker compose up --build` from the repository folder. 4. Open the studio in the browser and confirm the gallery lists your ComfyUI outputs. ### Prompt to paste into your agent ```text Clone github.com/tristmeister/HEISS-UI, set COMFY_OUTPUT_HOST to , and run `docker compose up --build`. Report the URL the studio is served on and whether the gallery lists existing ComfyUI outputs. Verify that the container is running and that the models list fills in, which means ComfyUI is reachable. ``` ### Caveat The README points to DOCKER.md for the reasoning behind the volume and folder setup and for a plain docker run, so read that guide before changing the mounts. Dual-use notice: only use this on software, sites and systems you own or are authorized to test. Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/ URL: https://dedupedai.com/use-cases/tristmeister-heiss-ui/run-the-studio-in-docker-next-to-comfyui/