boogu / base / createImage

Boogu-Image came out of nowhere in mid-2026 and landed a real first impression: the community's launch-week verdict was "same league or slightly superior to Klein," with strong understanding of public characters and artistic styles and a fast turbo demo. Civitai Image Gen (boogu / base / createImage) is that model's text-to-image node on Civitai's cloud - a chance to run a genuinely good 2026 base without downloading a thing.

What it is

The base (non-turbo) text-to-image operation for the Boogu ecosystem. Required: prompt. Optional: output_format, image_metadata, negative_prompt, sampler/scheduler, steps (default 20), cfg_scale (default 4), seed, quantity (1–12), loras, diffusion_model (CIVITAI_AIR socket, if you want to pin a specific Boogu checkpoint), width/height (64–2048), api_config.

Outputs: images (native IMAGE), errors, workflow_id, raw_json.

What the community actually found

The honest read from the early threads: Boogu is fast and clean, understands public characters decently, and handles artistic styles well. The criticisms were equally specific - some found the output "too synthetic" with low variability between seeds, and the semirealistic look got the standard "slop" dismissal from the usual quarter. Comfy Org took it seriously enough to ship ConvRot INT8 conversions of Boogu-Image, which is a real signal it entered the ecosystem's working roster.

So: it's a credible modern base, not a miracle. On the cloud it's priced per call, which makes it an easy model to try and an easy model to skip if it's not your aesthetic.

Install & gotchas

Standard pack install - ComfyUI Manager (search Civitai Comfy Nodes), comfy node registry-install civitai-comfy-nodes, or git clone + pip install -r requirements.txt. Auth via CIVITAI_API_TOKEN, OAuth, or a wired Civitai Auth node.

Worth knowing:

1. No ControlNet socket on the Boogu surface. If you need guided structure, this engine's API doesn't expose it - wire to a different ecosystem's node.
2. quantity up to 12, all billed. Every image is a Buzz charge; raw_json shows the per-run cost if you want proof.
3. Your prompt is the main lever. Boogu's character and style understanding is its strength, so describe the character/public figure and the style explicitly - that's where the model does its best work.

One honest take: Boogu is a "try it, see if it's your thing" model, and this node is the cheapest way to run that experiment - no GGUF download, no VRAM, no install beyond the pack. If the synthetic-look complaint was going to bother you, it'll bother you here too; the base's job is to show you the model, not to change your taste.