Openimage

Openimage is a terminal UI for generating AI images. Run it with no arguments to open the interactive TUI — type a prompt, watch a live progress bar, and see your image rendered inline. It also supports a CLI generate subcommand for scripting and one-liners.

It supports thirteen providers — OpenAI, OpenRouter, Stability AI, Replicate, Ideogram, DeepAI, GetImg.ai, Clipdrop, Segmind, Zhipu, Baidu, ByteDance, and Alibaba — as well as any OpenAI-compatible endpoint.

openimage constructs the appropriate API request for the chosen provider, authenticates with your credentials, downloads the resulting image, saves it to disk, and renders it inline in the terminal. All parameters are configurable via a YAML file or per-invocation flags.

---

Installation

go install github.com/anomalyco/openimage@latest

Or from source:

git clone https://github.com/anomalyco/openimage.git
cd openimage
go build -o openimage .

---

Quick start

openimage config set api-key sk-or-...
openimage

That opens the TUI. Type a prompt and press enter. The image generates with a live progress bar and renders inline.

For CLI usage:

openimage generate "a watercolor painting of mountains at sunset"
openimage generate "a cat" --enhance
openimage generate "a fantasy landscape" --iterate

---

Terminal UI

Running openimage with no arguments launches a full-screen interactive TUI:

┌─────────────────────────────────────────────────────┐
│                   ✧  openimage                      │
│                                                     │
│  describe the image you want to generate             │
│  ┌───────────────────────────────────────────────┐   │
│  │ > a cinematic portrait of a robot reading...  │   │
│  └───────────────────────────────────────────────┘   │
│                                                     │
│           enter  generate    ctrl+c  quit            │
└─────────────────────────────────────────────────────┘

Features:

• Live progress bar with gradient animation while generating
• Inline image rendering using Kitty, iTerm2, Sixel, or braille protocols
• Dark theme with accent colors and rounded borders
• Keyboard-driven: enter to generate, ctrl+c / q to quit

After generation, press enter to start a new prompt.

CLI Usage

openimage generate <prompt> [flags]

openimage generate "a robot reading a book"
openimage generate -p "a robot reading a book"
openimage generate "cyberpunk city" --provider stability -m sd3
openimage generate "logo concept" -o logo.png -s 1024x1024 --open
openimage generate "variations" -n 4 --provider replicate -m black-forest-labs/flux-schnell
openimage generate "prompt" --provider openai --api-base https://api.together.ai/v1 -m black-forest-labs/FLUX.1-schnell

Flags

Flag	Short	Description
--prompt	-p	Text prompt
--provider		Provider name
--model	-m	Model ID
--size	-s	Dimensions, e.g. 1024x1024
--quality	-q	Quality or speed tier
--style		Style preset
--output	-o	Output file path
--count	-n	Number of images (default 1)
--enhance		Improve prompt with an LLM before generation
--iterate	-i	Interactive generation with feedback and refinement
--api-base		Override API base URL
--api-key		Override API key
--open		Open result with system viewer

When --output is omitted, files are saved to save_dir with a timestamp: openimage-20260123-143052.png.

---

Configuration

openimage reads openimage.yaml from the current working directory. Every config value can be overridden by a command-line flag.

api_key: sk-or-v1-abc123
provider: openrouter
model: openai/dall-e-3
size: 1024x1024
save_dir: ~/Pictures/openimage
display: true

Key	Type	Default	Description
api_key	string	—	API key
api_base	string	—	Override the default API base URL
provider	string	openrouter	Provider name
model	string	openai/dall-e-3	Model ID
size	string	1024x1024	Image dimensions
quality	string	—	Quality or speed tier (provider-specific)
style	string	—	Style preset (provider-specific)
save_dir	string	.	Output directory (supports ~)
display	bool	true	Show images in the terminal

Manage config with subcommands:

openimage config set <key> <value>
openimage config get <key>
openimage config list
openimage config path

API key

Resolved in this order:

1. --api-key flag
2. OPENIMAGE_API_KEY env var
3. OPENROUTER_API_KEY env var (compatible with the OpenRouter CLI)
4. api_key in openimage.yaml

---

Providers

Name	Docs
openrouter	docs
openai	docs
stability	docs
replicate	docs
ideogram	docs
deepai	docs
getimg	docs
clipdrop	docs
segmind	docs
zhipu	docs
baidu	docs
doubao	docs
dashscope	docs

Any OpenAI-compatible endpoint works via --provider openai --api-base <url>.

---

Models

Model IDs are provider-specific. The table below lists some of the most popular ones. For each provider's full catalog, see the docs linked above.

Model	Provider	Notes
openai/dall-e-3	openrouter, openai	1024x1024, 1024x1792, 1792x1024
openai/dall-e-2	openrouter, openai	256x256, 512x512, 1024x1024
stability-ai/stable-diffusion-3	openrouter	SD3 via OpenRouter
stability-ai/stable-diffusion-xl	openrouter	SDXL via OpenRouter
black-forest-labs/flux-schnell	openrouter, replicate	Fastest Flux
black-forest-labs/flux-dev	openrouter, replicate	Flux Dev
black-forest-labs/flux-pro	openrouter, replicate	Flux Pro
bytedance/seedream-4.0	openrouter	Seedream 4 via OpenRouter
ideogram-ai/ideogram-v2	openrouter	Ideogram v2 via OpenRouter
google/gemini-2.5-flash-image	openrouter	Gemini 2.5 Flash, a.k.a. Nano Banana
google/gemini-3.1-flash-image	openrouter	Nano Banana 2 — Gemini 3.1 Flash Image
google/gemini-3-pro-image	openrouter	Nano Banana Pro — Gemini 3 Pro Image
google/gemini-3.1-flash-image-preview	openrouter	Nano Banana 2 (preview)
google/gemini-3-pro-image-preview	openrouter	Nano Banana Pro (preview)
openai/gpt-5-image	openrouter	GPT-5 + GPT Image 1
openai/gpt-5-image-mini	openrouter	GPT-5 Mini + GPT Image 1 Mini
openai/gpt-5.4-image-2	openrouter	GPT-5.4 + GPT Image 2
openrouter/auto	openrouter	Auto-routes to best model
gpt-image-2	segmind	GPT-4o image generation
flux-schnell	segmind, getimg	Fast Flux
flux-dev	segmind	Flux Dev
sdxl1.0-newreality-lightning	segmind	Fast SDXL
juggernaut-xl	segmind	Photorealistic SDXL
realistic-vision-v5	segmind	Photorealistic
dreamshaper-xl	segmind	Artistic SDXL
nano-banana-2	segmind	Fast general-purpose
sd3	stability	Stable Diffusion 3
sd3.5-large-turbo	segmind	SD3.5 Turbo
core	stability	Stable Image Core
ultra	stability	Stable Image Ultra
v4	ideogram	Latest, excellent text rendering
v3	ideogram	Previous version
seedream-5-lite	getimg	Fast and capable
seedream-4.0	getimg	Previous version
stable-diffusion-xl	getimg	SDXL
cogview-3-plus	zhipu	Highest quality CogView
cogview-3-flash	zhipu	Fast CogView
cogview-4	zhipu	Latest, multi-reference
sd_xl	baidu	ERNIE-ViLG SDXL
doubao-seedream-3.0-t2i-250415	doubao	Seedream 3.0 (ByteDance)
wan2.6-t2i	dashscope	Alibaba Wanxiang 2.6
wan2.5-t2i-preview	dashscope	Alibaba Wanxiang 2.5
kling/kling-v3-image-generation	dashscope	Kling V3 (Kuaishou)

---

Terminal display

openimage renders generated images directly in your terminal using the best available protocol, detected at runtime:

Protocol	Terminals
Kitty	Kitty
iTerm2	iTerm2, WezTerm, VS Code
Sixel	xterm, foot, contour, tmux (with sixel enabled)
Braille	All terminals — bash, zsh, etc. (Unicode braille characters)

The protocol is automatically selected based on your terminal environment ($TERM, $TERM_PROGRAM, $KITTY_WINDOW_ID). No configuration needed.

Disable inline display with openimage config set display false.

---

Development

Requires Go 1.22+.

git clone https://github.com/anomalyco/openimage.git
cd openimage
go mod tidy
go run .
go vet ./...
go build -o openimage .

cmd/
  root.go        Root command — launches TUI by default
  generate.go    generate subcommand (CLI)
  config.go      config subcommand
  tui.go         Terminal UI (bubbletea)
pkg/
  config/        YAML config read/write
  display/       Terminal image rendering (Kitty, iTerm2, Sixel, braille)
  provider/      Provider interface and implementations
main.go          Entry point

Adding a provider

1. Create pkg/provider/<name>.go implementing the Provider interface
2. Register it in New() in pkg/provider/provider.go
3. Add it to the help text in cmd/config.go and cmd/generate.go

Keep provider files self-contained, follow existing error handling patterns, and avoid new dependencies unless necessary.

---

Contributing

Bug reports, feature requests, and pull requests are welcome. Fork the repo, create a branch, run go vet ./..., and open a PR.

---

License

MIT — see LICENSE.md.