Innovate Futures @ Benji
patreon
(Workflow) LTX V2V Flow Edit With DeepSeek Janus Pro - Ver 20250130
🕑 Added 2025-01-30 14:28:43 +0000 UTC
This is an updated version of ltx_v2v-flow_edit workflow.
In this workflow, I swtiched from Florenece 2 to DeepSeek AI Janus Pro for vision LM and using prompting for style transfer.
Video Tutorial : https://youtu.be/66XSSH6jThs
Other JanusPro workflow in tutorial : https://www.patreon.com/posts/comfyui-janus-ai-121111621
Have Fun! And Keep Inspire!
Comments
Benjamin Law
For image descriptive, yes. At least better than general captioning. For object segment, no , its still Yolo segmentation better, or should say there's no segment ation in this VLM.
Clayfacer
interesting is Janus Pro better ?
