max

Making good software for generative media @tryremora

[max @passionandfury]

@andrewxroas Andrew open up your DMs so I can give you credits for @tryremora and get you off the slop interfaces you’re using for generation :)

[max @passionandfury]

@sonny8s @eptwts For seedance native audio generation is the move. Come FAFO on @tryremora :)

[max @passionandfury]

For what it's worth, I generated four pieces of output for each prompt and model and took the best one for comparisons sake in order to eliminate one-off issues. Made easy w/ @tryremora :)

[max @passionandfury]

Spent some time comparing Seedance 2.0 and Flux 3 over a few generations (comparisons below).

TLDR: Seedance 2.0 is still king for realism. Flux 3's output has a certain AI "smell" about it when it comes to object textures and facial features that Seedance 2.0 usually nails. With that said, Flux 3 is much more permissive with prompts as Seedance 2.0 still clamps down hard on human faces.

[max @passionandfury]

I wanted to also test how each model fared on animated output. I think Flux 3 really closed the gap here but the physics at the end are still off. Seedance 2.0 also has a physics issue with the berry at the end but much less pronounced than the floating squirrel on Flux 3

[max @passionandfury]

Flux 3 from @bfl_ai is now live on Remora.

Not only does the output rival seedance 2.5, but the real killer use case is the concept of drafts.

Why give a shit? Because drafts cost 79.3% less and are 44.4% faster to generate than full renders. And the real magic is that you can upscale a draft to a full render while MAINTAINING consistency.

If I had a dollar for every 480p seedance generation that looked perfect, and then used the same prompt/settings with 1080p only for it to look completely different...

[max @passionandfury]

Hi David. We're working on a tool for power users like yourself at @tryremora. While we might not mirror your workflow 1:1, we do offer really competitive pricing on Seedance 2.0 (~40% cheaper than Fal).

If you're ever open to giving us a shot, we'd love to learn from your workflow!

[max @passionandfury]

@oi_sady Give us a shot @tryremora will ya?

[max @passionandfury]

@rakhul @RoundtableSpace Say more? Why not just create another generation? What's there to undo?

[max @passionandfury]

@SaintandSpecter Curious which tool your using for generation and what your pain points are? Working on making a great generative media application and could use all the feedback I can get :)

[max @passionandfury]

@iGreenGod @Lianaalane @wavespeed_ai Is this issue happening across different models or does this seemed confined to a specific tool? Regardless I'm confident @tryremora could be of use here - our goal is to be the platform with the best UX

[max @passionandfury]

@o7_JustJames @WarriorMiMi17 Do you think this is something @tryremora can help with? We genuinely aim to be the best generative media tool out there - would love your feedback 🙏

[max @passionandfury]

@Weinbach @tec_aryan What are you trying to edit after generation? Would love to get a better sense of your use-case here - if we don't already have it in @tryremora would love to add it in

[max @passionandfury]

Make Loading Skeletons Great Again @tryremora

[max @passionandfury]

@danieldmai still happening? every output I get from seedance looks overly saturated, not sure what's up

[max @passionandfury]

@Tarunn_b @zeeg I'd switch to windows if it wasn't for https://t.co/PVPpC9oNEp

[max @passionandfury]

@andrewxroas While I have you, do you start campaigns with any objectives other than purchase to build data or do you just start off with purchase?