Unverified50% confidenceFactExact time
MiniDog's fine-tuning uses 2,000 images generated by FLUX.1-schnell, constituting fine-tuning on synthetic data.
1
Sources
50%
Confidence
Long-term
Relevance
10/14/2026
First Seen
Sources
Related Entities
Related Claims
Unverified微缩模型摄影（Miniature Faking）是一种通过移轴镜头或后期处理让真实场景看起来像玩具模型的摄影技法，核心在于制造浅景深效果、高饱和度色彩以及俯视视角61% similarUnverifiedMiniDog's pre-training uses 26,000 dog images from ImageNet, and generates text descriptions for them.61% similarUnverifiedMiniMax H3 REF outputs natively generated large images directly from the model, rather than pseudo-high-definition images produced through post-processing upscaling or stitching.57% similarUnverifiedThis creator uses the MiniMax Ref standard workflow, using original film screenshots as reference material for characters and scenes.57% similarUnverifiedMinimax is a multimodal large model from ByteDance, and H3 represents its image generation version.56% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1026753API
curl https://kongchang.com/api/v1/knowledge/claims/1026753MCP
get_claim(id=1026753)