This is a very simple workflow set that helps you to save your Ideogram 4.0 Text to Image Generation Data (Your final Ideogram 4.0 JSON Prompt) into a human readable .txt file. This will automatically get and write your image generation prompt to the .txt file. You will find all the saved prompt files that it generated with the images inside the Archive (.Zip) that has the workflow. Also with the Image Saver Simple node used inside the main workflow you may embed that workflow itself with each saved image or save the image and workflow for your work separately. As all the ComfyUI workflow are just JSON file I opted to save the Ideogram 4.0 prompt with .txt file extension to avoid confusion.. if you want to feed a prompt you like very much to other tools that accept .json files directly for image generation you can just copy the .txt file elsewhere on your system and change it's file extension to .json.
Ideogram 4.0 is an advanced text-to-image AI model developed by Ideogram — a company founded in 2023 by prominent researchers from Google Brain and the University of Toronto. A key differentiator of this model is its support for JSON-structured prompting. This capability allows users to pass highly specific, programmatic instructions detailing object placement, color palettes, and precise text rendering within the image. Furthermore, Ideogram 4.0 featured day-one integration with ComfyUI, a popular node-based interface for AI workflows. This immediate compatibility ensures that creators can seamlessly incorporate the model into their existing production pipelines.
Ideogram 4.0 is not very noob friendly and you would need some prior experience of ComfyUI to get good results, I tried my best to make the process of using it as easy, as compact and as memory efficint as possible so it may run on systems with low VRAM (6 or 8 GB VRAM). The default ComfyUI workflow is not compact and efficient so I hope mine will work well for you. I splitted the default workflow to two completely seperate .JSON files -
[Workflow 1] Ideogram4_Txt2JSON
-------------------------------
* Purpose: Acts as a local pre-processor to turn natural language into structural JSON layout schemas without loading the massive Ideogram diffusion model.
* How it works: A quantized vision-language model (Qwen3-VL-2B-Instruct running in a low VRAM-friendly 4-bit mode) interprets your text prompt and decomposes it into spatial bounding boxes.
* Output: Renders a structured JSON document, generates a visual canvas box layout via Kijai's Prompt Builder, and auto-saves the text template to your ComfyUI output directory as: "Ideogram4_Txt2JSON_[timestamp].txt".
[Workflow 2] Ideogram4_T2I
--------------------------
* Purpose: Loads the optimized FP4/NF4 mixed precision Ideogram weights and handles the heavy VRAM math to actually sample and decode the image safely.
* How it works: Imports the structural bounding boxes, allows LoRA injections (e.g., 80s Anime filters), processes through an AuraFlow-adapted sampler, and immediately runs automated aggressive RAM/VRAM cache-clearing functions ("VRAMCleanup" and "RAMCleanup") after saving the image to protect your system's memory headroom.
* Output: Images and raw text parameters are written directly to your output directory using the naming structure: Ideogram4_T2I_[timestamp].png.
You can download your necessary model file used in this workflow from HuggingFace and LORA from CivitAI (Details are mentioned below). Make sure you have latest enough ComfyUI installation and install any necessary nodes for for this workflow using ComfyUI manager and place the correct files in correct places. Also check out my other workflows for SD 1.5 + SDXL 1.0, Pony, WAN 2.1, WAN 2.2, MagicWAN Image v2, QWEN, HunyuanImage-2.1, HiDream E, HiDream O1, Ernie Image, KREA, Chroma, AuraFlow, NoobAI, Illustrious, Lumina2, Z-Image Turbo, Flux.2 Klein 9B & 4B, Flux.1 Dev and Kandinsky Image 5 Lite (T2I & I2I) models. Feel free to toss some yellow Buzz on stuffs you like.
@@@@@@@@@@@@@@@@@@@@@
Note that LORA usage for this workflow is optional you can use it without any LORAs, use with 1 or 2 or any other number of LORAs, to add new LORAs press the L button on top to lunch LORA Manager on a new tab find your LORA and if you want to use that LORA just click the upward Kite button. Also note If you see "Image blocked by safety filter" it is because of safety training in the model itself, ComfyUI does not have any safety filter. Some LORA usage with cleaver prompting (with Kijai's Prompt Builder) may give you your sesired NSFW results but completely bypassing the safty filters of Ideogram 4.0 is not always possible. I have included a mini how to for Kijai's Prompt Builder with this workflow that you may see.
Required Files
===============
### Download Link for Ideogram 4.0 models -
++++++++++++++++++++++++++++++++++++++++++++
This HuggingFace page has all the Ideogram 4.0 model (fp8 & nvfp4) files -
https://huggingface.co/Comfy-Org/Ideogram-4/tree/main
I have used the nvfp4 safetensors files for the image generation workflow which is perfect for any 8GB VRAM GPU, if you have 16 GB or higher VRAM GPU you may use the bigger fp8 files. For the Text to JSON Prompt builder you don't need to download any model, it all dependencies of that workflow are satisfied it's QwenVL Node will automatically download and manage it's own QWEN model.
LORAs Tested with this workflow -
++++++++++++++++++++++++++++++++++
Vintage Anime - Ideogram 4 -
https://civarchive.com/models/2685958/vintage-anime-ideogram-4?modelVersionId=3017988
zjourney - Fantasy Realism Refiner Ideogram 4 -
Vintage Beauties NEW - Ideogram 4 -
https://civarchive.com/models/940109/vintage-beauties-new-ideogram-4?modelVersionId=3010189
Realism Engine Ideogram 4 -
https://civarchive.com/models/2688234/realism-engine-ideogram-4?modelVersionId=3033241
The last two listed LORAs are capable of delivering some mouth watering NSFW generations if the safty filter is successfully bypassed