Skip to content

UPSTREAM PR #17577: model: LFM2-VL fixes #350

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
wants to merge 13 commits into
base: main
Choose a base branch
from
Open

UPSTREAM PR #17577: model: LFM2-VL fixes #350

Changes from 1 commit
Commits
Show all changes
13 commits
Select commit Hold shift + click to select a range
2386891
Adjust to pytorch
tdakhran Nov 25, 2025
c509073
Add antialiasing upscale
tdakhran Nov 27, 2025
80b4e97
Increase number of patches to 1024
tdakhran Nov 27, 2025
1cd4e2f
Handle default marker insertion for LFM2
tdakhran Nov 27, 2025
40e08b8
Switch to flag
tdakhran Nov 27, 2025
65789e5
Reformat
tdakhran Nov 28, 2025
7cf67d6
Cuda implementation of antialias kernel
tdakhran Nov 28, 2025
7c8b098
Change placement in ops.cpp
tdakhran Nov 28, 2025
3ea706e
consistent float literals
tdakhran Nov 28, 2025
0b14906
Pad only for LFM2
tdakhran Nov 28, 2025
b81928f
Address PR feedback
tdakhran Nov 30, 2025
31be1a9
Rollback default marker placement changes
tdakhran Nov 30, 2025
2385ecf
Fallback to CPU implementation for antialias implementation of upscale
tdakhran Nov 30, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
Prev Previous commit
Next Next commit
Pad only for LFM2
  • Loading branch information
@tdakhran
tdakhran committed Nov 30, 2025
commit 0b14906ab7b39039dbbc61de39e58a5fa3cd21e2
4 changes: 3 additions & 1 deletion tools/mtmd/clip.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -4355,9 +4355,11 @@ bool clip_image_preprocess(struct clip_ctx * ctx, const clip_image_u8 * img, str
params.patch_size * params.n_merge,
params.image_min_pixels,
params.image_max_pixels);
const std::array<uint8_t, 3> pad_color = {122, 116, 104};

clip_image_u8 resized_img;
img_tool::resize(*img, resized_img, target_size, img_tool::RESIZE_ALGO_BILINEAR, false);
const bool pad = (ctx->proj_type() != PROJECTOR_TYPE_LFM2);
img_tool::resize(*img, resized_img, target_size, img_tool::RESIZE_ALGO_BILINEAR, pad, pad_color);
clip_image_f32_ptr res(clip_image_f32_init());
normalize_image_u8_to_f32(resized_img, *res, params.image_mean, params.image_std);
res_imgs->entries.push_back(std::move(res));
Expand Down