Nodes/civitai-comfy-nodes/boogu / turbo / createImage
ComfyUI Node

boogu / turbo / createImage

Boogu Turbo — the 4-step draft machine, cloud-served

By civitai·Created 2 months ago·Updated about a month ago· 42
boogu / turbo / createImage
  • loras
  • diffusion_model
  • api_config
  • images
  • errors
  • workflow_id
  • raw_json
prompt
output_format
image_metadata
negative_prompt
sampler
scheduler
steps4
cfg_scale1.00
seed0
quantity1
width1024
height1024

Boogu's demo was a turbo demo, and that's what made people notice it - four steps, guidance off, images that were "very clean and without too many mistakes" for a model nobody saw coming. Civitai Image Gen (boogu / turbo / createImage) is that distilled variant as a cloud node: the fastest way to hit Boogu's output on a machine with no GPU.

What it is

The turbo text-to-image operation for the Boogu ecosystem. It's the same node shape as the base Boogu create node with the sampling defaults rewritten for distillation: steps defaults to 4 (vs 20) and cfg_scale to 1.0 (vs 4). Required: prompt. Optional: output_format, image_metadata, negative_prompt, sampler/scheduler, seed, quantity (1–12), loras, diffusion_model (CIVITAI_AIR), width/height, api_config.

Outputs: images, errors, workflow_id, raw_json.

How it fits

The turbo/base split is the standard 2026 deal: the distilled model converges in a handful of steps and runs guidance-off, trading the base's headroom for speed. On the cloud that trade is doubly visible - speed is literally money, since you're paying per call. The community's first-impressions thread praised turbo's cleanliness and character understanding specifically, while also flagging low seed-to-seed variability and a slightly synthetic look. Those are the model's, not the node's.

The one real trap: don't treat 4 steps as a starting point to raise. The model is distilled for 4; cranking steps toward 50 fights the distillation and buys you nothing. If you want the base's ceiling, use the boogu / base / createImage node.

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.

And the standard cloud notes: every image is billed in Buzz (quantity up to 12 multiplies it), there's no ControlNet socket on the Boogu surface, and your prompt is the main lever - Boogu's strength is understanding characters and artistic styles, so say both explicitly.

Honest take: if you're going to run Boogu at all, turbo is the sensible first stop - it's the version the community actually tested and praised for speed, and on the cloud "fast" also means "cheap." Start here, and only step up to the base if the draft quality isn't landing.

CategoryCivitai/Image/boogu

Inputs (15)

NameTypeDefaultDescription
promptSTRING
output_formatoptCOMBO4 options: , jpeg, png, webP
image_metadataoptSTRINGExternal metadata that will be stored with the image
negative_promptoptSTRING
sampleroptCOMBO32 options: , euler, euler_ancestral, euler_cfg_pp, euler_ancestral_cfg_pp, heun, +26
scheduleroptCOMBO8 options: , normal, karras, exponential, sgm_uniform, simple, +2
stepsoptINT41–150
cfg_scaleoptFLOAT1.000–30
seedoptINT00–4294967295
quantityoptINT11–12
lorasoptCIVITAI_LORAS
diffusion_modeloptCIVITAI_AIR
widthoptINT102464–2048
heightoptINT102464–2048
api_configoptCIVITAI_CONFIGOptional Civitai Auth connection; defaults to CIVITAI_API_TOKEN or stored OAuth login.

Outputs (4)

NameTypeDescription
imagesIMAGE
errorsSTRING
workflow_idSTRING
raw_jsonSTRING