Image Stage

Generates one or more images from text with local or built-in models—the main “create from scratch” entry in ComfyTV, with img2img (i2i) support.

Image Stage · screenshot coming soon

What this node does

Image Stage handles text-to-image (t2i) and image-to-image (i2i). Write main_prompt, pick a workflow (e.g. Local SD1.5 or Ideogram4), set resolution, aspect ratio, and batch size, click ▶ Run, and thumbnails appear on the node.

Two outputs:

• images (COMFYTV_IMAGES): the full batch from this run.
• image (COMFYTV_IMAGE): the selected thumbnail (selected_index, 1-based).

On the first successful run, ComfyTV usually auto-spawns an Image Picker downstream to pick one image for crop, upscale, or inpaint.

When to use it

• Generate concept art, product shots, or character designs from prompts.
• i2i: choose an i2i workflow, wire a reference to images, describe the desired change in main_prompt.
• Batch 2–8 images and compare compositions before editing one.
• Wire Text Stage text → texts for LLM-expanded prompts.
• Feed Video Stage I2V / FLF2V with image (COMFYTV_IMAGE).

How ComfyTV stages work

• ▶ Run executes only Image Stage’s subgraph—not the whole ComfyUI graph.
• Snapshots: image URLs are stored in the project; downstream Upscale/Crop reads them without re-running Image Stage.
• The workflow dropdown maps to JSON in workflows/image/; ComfyTV maps prompt, resolution, aspect_ratio, batch_size, and seed into EmptyLatentImage, KSampler, CLIPTextEncode, etc.

i2i workflows bind the first upstream images entry to LoadImage in the subgraph.

Types (COMFYTV_IMAGE vs ComfyUI IMAGE)

ComfyTV type	What it is	vs native ComfyUI
COMFYTV_IMAGE	Single-image project snapshot (/view? URL)	Not an in-memory IMAGE tensor
COMFYTV_IMAGES	Batch JSON {images:[...]}	Not a native IMAGE batch
Native IMAGE	Pixel tensor in GPU memory	Used by SaveImage; cannot wire directly to ComfyTV stages

Conversion:

• Native → ComfyTV: ComfyTV/Bridge → → ComfyTV Image (Run to snapshot)
• ComfyTV → native: ← ComfyTV Image (snapshot → IMAGE tensor for ControlNet, IPAdapter, etc.)

See bridges.md.

Common mistake: Wiring native Load Image (IMAGE) to images—types do not match. Use → ComfyTV Image Bridge or Load Image from Asset (COMFYTV_IMAGE).

Parameters

workflow

Backends in workflows/image/:

Option	Mode	Notes
Local SD1.5	t2i	Classic SD1.5 text-to-image
Local SD1.5 I2I	i2i	Same checkpoint, VAEEncode + denoise < 1
Image Ideogram4 T2I	t2i	Ideogram 4 + Qwen3-VL text encoder

Models: models.md. Details: README.md.

main_prompt

Describe the scene. Upstream texts append as extra context.

Example: a red apple on a wooden table, soft window light, photorealistic

resolution

Short-side tier: 480P, 720P, 1K, 1080P, etc. Combined with aspect_ratio for (w, h) on the latent node.

aspect_ratio

e.g. 1:1, 16:9, 9:16. Use 9:16 for vertical posters, 16:9 for widescreen.

batch_size

Images per run (1–8). image output uses selected_index.

texts

Wire Text Stage text or other COMFYTV_TEXT for extra prompt context.

images

Required for i2i: pick Local SD1.5 I2I, wire reference to images.image0 (first image wins). t2i workflows ignore upstream images.

Sources: image from Image Stage/Picker, Asset Loader, Bridge → ComfyTV Image.

selected_index

Which batch item image refers to (1-based). Click thumbnails on the node to switch; toolbar appears (✏️ Edit, 🌐 Panorama, etc.).

custom_params

Sidebar bindings for seed, negative prompt, steps, CFG, etc.

Outputs

Output	Type	Meaning	Downstream
images	COMFYTV_IMAGES	Full batch	Image Picker, Compare Stage
image	COMFYTV_IMAGE	Selected single	Video I2V, Upscale, Inpaint, Bridge ← ComfyTV Image

Step by step for beginners

1. Add Project, then ComfyTV → Generate → Image Stage.
2. workflow → Local SD1.5.
3. main_prompt: a red apple on a wooden table.
4. resolution 1K, aspect_ratio 1:1, batch_size 1.
5. Download v1-5-pruned-emaonly.safetensors (models.md).
6. ▶ Run; Image Picker may appear automatically.
7. Click a thumbnail → Upscale or Crop from the toolbar.

i2i: Local SD1.5 I2I → wire image to images → prompt e.g. watercolor style → Run.

Full guides (recommended reading)

This page covers one node only. For end-to-end workflows, multi-stage pipelines, type conversion, and design rationale, see the ComfyTV user guides on GitHub:

Guide	Contents
Getting started	Install, canvas basics, per-stage Run, snapshots, Project, Image Picker
Generate	Text, Image, Video, Music, and Speech stages and workflow selection
Image tools	Crop, inpaint, outpaint, upscale, multi-angle, variation presets
Model files	Checkpoints, LoRAs, and folder paths per workflow
Custom workflows	Import your ComfyUI workflow JSON without Python changes

Repository and workflows

Resource	Link
GitHub repository	https://github.com/jtydhr88/ComfyTV
User guides index	https://github.com/jtydhr88/ComfyTV/tree/main/docs
Built-in workflows	https://github.com/jtydhr88/ComfyTV/tree/main/workflows
Model checklist	https://github.com/jtydhr88/ComfyTV/blob/main/docs/models.md
This node’s workflow folder	https://github.com/jtydhr88/ComfyTV/tree/main/workflows/image
Workflow README	https://github.com/jtydhr88/ComfyTV/blob/main/workflows/image/README.md
Custom workflows	https://github.com/jtydhr88/ComfyTV/blob/main/docs/custom-workflows.md

FAQ

Q: Run does nothing or workflow is grey?
A: Check models; restart ComfyUI; read terminal errors.

Q: Cannot wire native Load Image to images?
A: Wrong type—use Bridge → ComfyTV Image or Load Image from Asset.

Q: Batch has 4 images but downstream gets one?
A: Use image or Image Picker; images is the full batch JSON.

Q: Changed prompt but downstream still shows old image?
A: Re-run Image Stage, then downstream stages.

Related nodes

• Image Picker—pick one from batch or accumulated pool (often auto-created).
• Text Stage—upstream prompt expansion.
• Video Stage (I2V / FLF2V)—image as video conditioning.
• Upscale / Inpaint / Outpaint Stage—single-image editing.
• Bridge → / ← ComfyTV Image—native ComfyUI interop.
• Project—project context.