Academia

This page is auto-generated from model configurations. Last updated: 2026-07-01.

This reference lists all available Academia image generation models and their parameters. Use these parameter names when calling the Generation API.

• ERNIE Image
• ERNIE Image Turbo
• Joyai Image Edit
• Physic Edit
• Telestyle V2 Style Transfer
• Z-Image
• Z-Image Turbo

---

ERNIE Image

Baidu ERNIE-Image (SFT) — strong instruction following, text rendering, and layout-aware generation; ~50 inference steps with optional prompt enhancer.

Model ID: model_ernie-image

Capabilities: txt2img

LLM Markdown: https://app.scenario.com/api/models/model_ernie-image/markdown

Parameter	Type	Required	Default	Min	Max	Allowed Values	Description
prompt	string	Yes	-	-	-	-	Text prompt for image generation.
numOutputs	number	No	1	1	4	-	Number of images to generate.
width	number	No	1024	64	2048	-	Width of the generated image. Recommended sizes are listed in presets.
height	number	No	1024	64	2048	-	Height of the generated image. Recommended sizes are listed in presets.
numInferenceSteps	number	No	50	1	100	-	Number of denoising steps. The SFT model is tuned for ~50 steps.
guidanceScale	number	No	4	0	20	-	Classifier-free guidance scale.
usePe	boolean	No	true	-	-	-	Expand short prompts into richer descriptions (recommended on the model card).
seed	number	No	-	-	-	-	Random seed for reproducible generation.

ERNIE Image Turbo

Baidu ERNIE-Image-Turbo — fast ~8-step generation with optional prompt enhancer (guidance fixed in the deployment).

Model ID: model_ernie-image-turbo

Capabilities: txt2img

LLM Markdown: https://app.scenario.com/api/models/model_ernie-image-turbo/markdown

Parameter	Type	Required	Default	Min	Max	Allowed Values	Description
prompt	string	Yes	-	-	-	-	Text prompt for image generation.
numOutputs	number	No	1	1	4	-	Number of images to generate.
width	number	No	1024	64	2048	-	Width of the generated image. Recommended sizes are listed in presets.
height	number	No	1024	64	2048	-	Height of the generated image. Recommended sizes are listed in presets.
usePe	boolean	No	true	-	-	-	Expand short prompts into richer descriptions (recommended on the model card).
seed	number	No	-	-	-	-	Random seed for reproducible generation.

Joyai Image Edit

All-in-one image editing model with strong visual understanding for precise natural language edits.

Model ID: model_joyai-image-edit

Capabilities: img2img

LLM Markdown: https://app.scenario.com/api/models/model_joyai-image-edit/markdown

Parameter	Type	Required	Default	Min	Max	Allowed Values	Description
prompt	string	Yes	-	-	-	-	The edit instruction describing what changes to make to the image.
image	file	Yes	-	-	-	-	Input image to edit.
negativePrompt	string	No	-	-	-	-	Text describing what should not appear in the output image.
numInferenceSteps	number	No	30	1	50	-	Number of denoising steps.
cfgScale	number	No	5	1	15	-	Controls how closely the output follows the edit instruction.
seed	number	No	-	0	2147483647	-	Random seed for reproducibility.

Physic Edit

Physics-aware image editing with realistic refraction, material changes, and deformations.

Model ID: model_physic-edit

Capabilities: img2img

LLM Markdown: https://app.scenario.com/api/models/model_physic-edit/markdown

Parameter	Type	Required	Default	Min	Max	Allowed Values	Description
image	file	Yes	-	-	-	-	Input image to edit with physics-aware transformations.
prompt	string	Yes	-	-	-	-	Editing instructions describing the physical transformation to apply.
negativePrompt	string	No	-	-	-	-	Description of unwanted elements in the output.
numInferenceSteps	number	No	40	1	50	-	Number of denoising steps.
cfgScale	number	No	4	1	20	-	Classifier-free guidance scale.
seed	number	No	-	0	2147483647	-	Random seed for reproducibility.

Telestyle V2 Style Transfer

Restyle an image with TeleStyle v2 by transferring style, material, lighting, and color from a reference while preserving the content image composition.

Model ID: model_telestyle-v2

Capabilities: img2img

LLM Markdown: https://app.scenario.com/api/models/model_telestyle-v2/markdown

Parameter	Type	Required	Default	Min	Max	Allowed Values	Description
contentImage	file	Yes	-	-	-	-	The image you want to restyle. Its subject and composition are kept; only the look changes.
styleImage	file	Yes	-	-	-	-	The image whose look you want to borrow — its artistic style, materials, lighting, and colors are applied to your content image.
resolution	string	No	square_hd	-	-	square_hd, square, portrait_4_3, portrait_16_9, landscape_4_3, landscape_16_9	The size and shape of the result.
loraScale	number	No	1	0	4	-	How strongly the reference style is applied. Higher values push the styling harder; lower values keep more of the original look.
useLightningLora	boolean	No	true	-	-	-	Adds a speed booster that generates results faster. On by default.
lightningLoraScale	number	No	1	0	4	-	How strongly the speed booster is applied. Only matters when “Use Lightning” is on.
useContentDescription	boolean	No	true	-	-	-	Automatically analyzes your content image and feeds a written description into the process, which can improve how well the subject is preserved.
useStyleDescription	boolean	No	true	-	-	-	Automatically analyzes your style image and feeds a written description into the process, which can improve how well the style is captured.
negativePrompt	string	No	-	-	-	-	Things to keep out of the result — for example, “text, watermark, blurry.”
numInferenceSteps	number	No	4	2	50	-	How many refinement passes the model makes. More steps can improve quality but take longer.
guidanceScale	number	No	1	0	20	-	How closely the result follows the model’s internal prompt. Higher values stick to it more strictly.
acceleration	string	No	none	-	-	none, regular	An optional speed boost. Regular is faster; None prioritizes quality.
numOutputs	number	No	1	1	4	-	How many images to create at once.
seed	number	No	-	0	2147483647	-	A number that makes results repeatable. Reusing the same seed and settings produces the same image; leave it empty for a different result each time.

Z-Image

Z-Image - Highest quality text-to-image and image-to-image generation

Model ID: model_z-image

Capabilities: txt2img, img2img, controlnet

LLM Markdown: https://app.scenario.com/api/models/model_z-image/markdown

Parameter	Type	Required	Default	Min	Max	Allowed Values	Description
modelId	model	No	-	-	-	-	A single LoRA model ID. If provided, it will be added to the loras array.
loras	model_array	No	“	-	-	-	List of one or more LoRA model IDs. If modelId is a LoRA and this is not provided, it will be automatically populated with the modelId.
lorasScale	number_array	No	-	0	2	-	Scales for the LoRA weights
prompt	string	Yes	-	-	-	-	Text prompt for image generation
negativePrompt	string	No	-	-	-	-	Text prompt for what to avoid in the image
image	file	No	-	-	-	-	Source image for img2img transformation
strength	number	No	0.6	0	1	-	How much to transform the reference image. 0 means no change, 1 means complete transformation
controlImage	file_array	No	-	-	-	-	Control images to use as input (edge, depth, or pose)
controlnetConditioningScale	number	No	0.75	0	1	-	The scale of the controlnet conditioning
numOutputs	number	No	4	1	4	-	Number of images to generate
numInferenceSteps	number	No	28	1	100	-	Number of inference steps.
width	number	No	-	128	2048	-	The width of the generated images, must be a multiple of 16
height	number	No	-	128	2048	-	The height of the generated images, must be a multiple of 16
guidance	number	No	5	1	10	-	Controls how closely the generated image follows the prompt
seed	number	No	-	0	2147483647	-	Used to reproduce previous results

Z-Image Turbo

Z-Image Turbo - Fast text-to-image and image-to-image generation model

Model ID: model_z-image-turbo

Capabilities: txt2img, img2img, controlnet

LLM Markdown: https://app.scenario.com/api/models/model_z-image-turbo/markdown

Parameter	Type	Required	Default	Min	Max	Allowed Values	Description
modelId	model	No	-	-	-	-	A single LoRA model ID. If provided, it will be added to the loras array.
loras	model_array	No	“	-	-	-	List of one or more LoRA model IDs. If modelId is a LoRA and this is not provided, it will be automatically populated with the modelId.
lorasScale	number_array	No	-	0	2	-	Scales for the LoRA weights
prompt	string	Yes	-	-	-	-	Text prompt for image generation
negativePrompt	string	No	-	-	-	-	Text prompt for what to avoid in the image
image	file	No	-	-	-	-	Source image for img2img transformation. Cannot be used with ControlNet.
strength	number	No	0.75	0	1	-	How much to transform the reference image. 0 means no change, 1 means complete transformation
controlImage	file_array	No	-	-	-	-	Control images to use as input (edge, depth, or pose). Cannot be used with Reference Image.
controlnetConditioningScale	number	No	0.7	0	1	-	The scale of the controlnet conditioning
numOutputs	number	No	4	1	4	-	Number of images to generate
numInferenceSteps	number	No	9	1	100	-	Number of inference steps.
width	number	No	-	128	2048	-	Width of the generated image
height	number	No	-	128	2048	-	Height of the generated image
seed	number	No	-	0	2147483647	-	Used to reproduce previous results