Realtime LCM tool on HackerShow

realtime image generation

SHOW HN2023

Shown by dvrp on Nov 13, 2023. Click the card to turn it over.

12points

12comments

#2,206of 3,109 in 2023

Top 71%of 2023 by points

How HN took it

QuietFewer than 5 comments about the project

0 positive, 4 neutral or asking, 0 mixed, 0 critical, out of 4 comments about the project.

Each row is a top-level comment and its replies, in HN's order. Select a cell to open that comment on HN.

positive neutral or mixed critical about another commenter or something else the maker

Fast inference times noted as impressive
From 1 comment
- Are y'all running the diffusion in PyTorch eager mode? AITemplate, stable-fast, or even a torch.compile could get that down to 60ms, I bet. Though I'm not sure the example…
  brucethemoose2 on HN

Claimed 50-100ms speeds may rely on quality-degrading methods
From 1 comment
- They might be "cheating" with methods that significantly degrade quality.
  brucethemoose2 on HN
Could use fewer inference steps elsewhere at quality cost
From 1 comment
- LCMs are actually exactly SD architecture. LCM is initialized from a regular SD unet and finetuned on a new objective. We are already compiling to get to these times. A lot of…
  erwannmillon on HN

Support for custom images
From 1 comment
- Can you use custom images?
  vmatsiiako on HN

Is it running in PyTorch eager mode?
From 1 comment
- Are y'all running the diffusion in PyTorch eager mode? AITemplate, stable-fast, or even a torch.compile could get that down to 60ms, I bet. Though I'm not sure the example…
  brucethemoose2 on HN
Is it running on H100s or A100s?
From 1 comment
- Is it running on H100s or A100s now?
  ilaksh on HN

stable-fast
From 2 comments
- > already compiling Hmm, well if you mean torch.compile, y'all should still check out stable-fast, which is claiming ~16ms/iter on a 4090:…
  brucethemoose2 on HN
- Are y'all running the diffusion in PyTorch eager mode? AITemplate, stable-fast, or even a torch.compile could get that down to 60ms, I bet. Though I'm not sure the example…
  brucethemoose2 on HN
AITemplate
From 1 comment
- Are y'all running the diffusion in PyTorch eager mode? AITemplate, stable-fast, or even a torch.compile could get that down to 60ms, I bet. Though I'm not sure the example…
  brucethemoose2 on HN