Fix WanVAE_.encode() to return both mu and log_var

[Mr-Neutr0n]

Summary

• Fixed WanVAE_.encode() to return both mu and log_var as expected by forward()
• Made scale parameter optional (defaults to None) in both encode() and decode() methods

Problem

The forward() method calls self.encode(x) expecting two return values (mu, log_var):

mu, log_var = self.encode(x)

However, encode() was only returning mu, causing a ValueError: not enough values to unpack.

Additionally, encode(x, scale) required the scale parameter, but forward() didn't pass it.

Changes

1. Modified encode() to return (mu, log_var) tuple
2. Made scale parameter optional with scale=None default
3. Made scale parameter optional in decode() as well for consistency
4. Updated WanVAE.encode() wrapper to extract only mu with [0] index

Related Issue

Fixes #556

Test Plan

• forward() now works without passing scale (uses no scaling when scale=None)
• Explicit scale parameter still works as before
• WanVAE wrapper continues to work unchanged

[Mr-Neutr0n]

bump on this — the encode() method was only returning mu and dropping log_var, which breaks anyone trying to use the VAE's full latent distribution. would appreciate a review

[Mr-Neutr0n]

Hi! Quick status: this still applies cleanly to the latest main as of today, and the PR is still mergeable. Re-requesting a review. If this looks good now, I would be happy to see it merged; if the repo has moved on, I am just as happy to close it.

[Mr-Neutr0n]

Closing this one to tidy up my open pull requests.

It's been open around five months with no review activity, and I'd rather withdraw it than leave it sitting in your queue. Nothing needed from you, and no hard feelings at all.

If the fix is still wanted, this can be reopened, or I'm happy to redo it properly against current main. Apologies for the noise.