Home | Notifications | New Note | Local | Federated | Search | Logout

Note Detail

Reply to @icedquinn@blob.cat
Q.U.I.N.N.@icedquinn@blob.cat (2026-10-16 07:37:39)
@strypey admittedly i'm the wrong person to ask because i constantly post about models that do work people genuinely are not interested in doing--like Cascadeur filling in tween frames between a human animator's keyframes, or CorridorKey perfectly green screening (a process that otherwise requires humans to sit there and manually trace out rotoscops and/or futz with despilling)

i would personally interact with humans more but they don't want to engage in any of the topics i do, nor do they want to participate in any of the coding projects i proposed over the years.

so in my cases its usually labor that *nobody* wants to do and isn't particularly creative, or just realistically would not have been made at all otherwise. notably a couple years ago i openly asked fedi repeatedly to draw some stream art for money and nobody would let me pay them to do a simple spritework task.
---Reply---
Strypey@strypey@mastodon.nzoss.nz (2026-10-16 10:03:07)
@icedquinn
> i don't know of any humans that are willing to voice act 8+ hours of an audiobook ... for free

We've had perfectly functional text>speech since the 1980s (I used S.A.M on a Commodore 64 FFS). So this comes under the category of tasks that can be done with more deterministic forms of AI.

Using a generative model is using an elephant gun to shoot a fly.

(1/?)

---Replies---

Blurry Moon@sun@shitposter.world (2026-10-16 10:14:38)
@strypey @icedquinn seriously fucking suggesting that SAM is suitable for narrating an audio book lol