Seedream5.0 Pro - Text to Image

Prompt

Pure hand-drawn 2D Japanese TV anime episode screencap, raw paused frame from a normal broadcast episode. [Saber Alter ] redesigned as traditional 2D anime characters, not game models. Clean ink lineart, flat anime paint, simple 2-tone/3-tone hard-edged shadows, warm skin tones, restrained colors, hand-painted background plate, 16:9 TV frame. Random mid-scene pause, natural imperfect poses, hands mid-gesture, mouths mid-syllable, eyes naturally tracking, not a key visual, not splash art, not a poster. Subtle 1080p broadcast softness, faint banding, tiny compression artifacts. Strictly no 3D, no 2.5D, no CGI, no cel-shaded game render, no Unreal/Unity/Blender/MMD/Vtuber look, no PBR, no glossy shaders, no plastic skin, no ambient occlusion, no bokeh, no HDR, no lens flare, no cinematic lighting. If it starts looking 3D, flatten it back into real 2D anime lineart and TV anime paint.

image

Seedream5.0 Pro - Text to Image

Seedream5.0 Proimage:seedream5-0-pro-text-to-imageImage generation
Aug 30, 2026, 08:46 PM

Settings

Provider
Seedream5.0 Pro
Model
image:seedream5-0-pro-text-to-image
Created
Aug 30, 2026, 08:46 PM
Credits
7
Mode
Image generation
Aspect ratio
21:9

Create Something Similar in GPTImage25

Use the generator below to remix the idea with the available model and settings.

Generate Image
New tasks appear here after you submit.