Will Weinbach

vibe coders be like…

Multica is solid for multi-agent delegation. Assign tasks directly instead of claiming. Where it gets powerful: agents autonomously creating and delegating tasks to peers with the right skills.

Higgsfield (@higgsfield) Soul 2.0 might be the most realistic image gen model to date, but it comes at a cost. Hallucinations are far more common and prompt adherence is limited.

What custom skills or workflows have you built for your Hermes or OpenClaw?

Here's the video generation workflow I keep coming back to for branded content: • Start with real product photos • Run them through Nano Banana 2 with a ChatGPT vibe prompt • Have ChatGPT write the video prompt from that output • Drop the image and prompt into Seedance 2.0

After spending time with Hermes, I can say its action transparency is a clear step above anything else on the market.

I ran 1,000 generations through Seedance 2.0. Here’s what actually matters: What it does well: • Strong with existing assets (image → video, reference → video) • Follows prompts better than expected • Audio context handling is ahead of most competitors Where it breaks: • Dialogue isn’t reliable. If timing is off, it starts hallucinating • Motion realism isn’t there yet. Moving objects still feel unnatural • No native 1080p / 2K / 4K. You’ll need to upscale everything

@Weinbach 😭😭😭😭😩😩😩😩😩😩😩

@Weinbach Here..

@Weinbach Now…who are you??