How to use Feta Enhance?

[rkfg]

I tried adding this node to the official LTX I2V workflow (0.9.1) (LTXV Loader => LTX Feta Enhance => LTXV Apply STG) and there's no effect at all. Changing the weight also doesn't have any effect. There are no examples in this repo either so I'm not sure if I'm doing something wrong.

[Skol600ml]

According to my tests, it works best without perturbed attention and in version 0.9. Connecting directly to the model. attn_override 8

You have to be careful when testing, to test with or without, you need to completely remove the model from the cache (manually), any change that alters the model itself without the model being reloaded is useless.

[Kwisss]

I did a double test and the node does nothing at all, output is 100% identical.

I obviously need to test some more, I did however get it to respond using the stg workflow and replacing the stg nodes with the feta ones

[logtd]

For that node you also need to use the Modify LTX Model node

[ApexArtist]

For that node you also need to use the Modify LTX Model node

so whats the correct sequence node flow ? load checkpoint, feta, modify model or modify model-feta

[phazei]

I ran it through GPT, it says that the Modify Model adds the classes necessary to be extended by the other checkpoints to modify. So Modify Model should come first, but after any attention or cache nodes.