joyai echo multishot support
this person merged joyai echo and ltx2.3, audio is fixed. What is missing is 'memory ref', can you implement a native comfyui node based on his worflow？ it will be a great contribution to the community. joyai echo is really like wan storymem, really good at multishot.
https://huggingface.co/joeygambino/joyai-echo-multishot-workflow
https://huggingface.co/joeygambino/joyai-echo-ltx23-echoVid-ltxAud-surgical-int8
Even without new custom node, surgical-int8 maintains more consistency in other shots.
The workflows examples for those above models are very complicated (in my view...), and also very resource heavy (being wrappers, not native)...
Yhere are already some attempts at simpler ways (by JoeyGamino as well as others.. and of course if Kijai wants/get the time, it would be great too).
I am making a workflow for this one, that seems to be much easier to use and works with native model loaders :
https://github.com/hashu786/comfyui-hrnodes/tree/main#-joyecho--multi-shot-av-memory-for-ltx-23
Seems to work well (using Joey Gaminos models)
(if you want to try already, there is a workflow in the repro from Hashu as well)
Early results (6 sequential videos) using the Hashu node (the first segment has an experimental I2V injection so that part might be better without) :
so cool , thanks. i think it is a game changer