We lifted 15 functions out of this paper's own repositories and ran 14 of them in a sandbox. "Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results.
| Repository | Role | Ran |
|---|---|---|
| zrealli/TIGIC | canonical | 14 of 15 |
| Function | Status | Where it lives |
|---|---|---|
| Normalize | Ran | zrealli/TIGIC/ldm/modules/diffusionmodules/model.py code served (permissive licence) · get_code("9fcdaa6e423e8aa7") |
| append_dims | Ran | zrealli/TIGIC/ldm/models/diffusion/sampling_util.py code served (permissive licence) · get_code("082e51fd548d78b2") |
| default | Ran | zrealli/TIGIC/ldm/modules/attention.py code served (permissive licence) · get_code("424012cb37b31172") |
| disabled_train | Ran | zrealli/TIGIC/ldm/models/diffusion/ddpm.py code served (permissive licence) · get_code("4cb732f513d69dfd") |
| exists | Ran | zrealli/TIGIC/ldm/modules/attention.py code served (permissive licence) · get_code("aa5486a3650902d8") |
| get_timestep_embedding | Ran | zrealli/TIGIC/ldm/modules/diffusionmodules/model.py code served (permissive licence) · get_code("cb49209c125de1b4") |
| isimage | Ran | zrealli/TIGIC/ldm/util.py code served (permissive licence) · get_code("b1368330cf0f5642") |
| ismap | Ran | zrealli/TIGIC/ldm/util.py code served (permissive licence) · get_code("d72762b700feee6f") |
| nonlinearity | Ran | zrealli/TIGIC/ldm/modules/diffusionmodules/model.py code served (permissive licence) · get_code("3137073275f8c21a") |
| norm_thresholding | Ran | zrealli/TIGIC/ldm/models/diffusion/sampling_util.py code served (permissive licence) · get_code("9898fe8b041f354c") |
| spatial_norm_thresholding | Ran | zrealli/TIGIC/ldm/models/diffusion/sampling_util.py code served (permissive licence) · get_code("d86694c13b94ae3a") |
| text_under_image | Ran | zrealli/TIGIC/ptp_scripts/ptp_utils_ori.py code served (permissive licence) · get_code("85d5eb123674597d") |
| uniform_on_device | Ran | zrealli/TIGIC/ldm/models/diffusion/ddpm.py code served (permissive licence) · get_code("d48d8354986e3b0e") |
| uniq | Ran | zrealli/TIGIC/ldm/modules/attention.py code served (permissive licence) · get_code("9a299fe5ae09e407") |
| log_txt_as_img | Not yet run | zrealli/TIGIC/ldm/util.py code served (permissive licence) · get_code("f9bd2e83191afad1") |
Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.
Despite significant advancements in image customization with diffusion models, current methods still have several limitations: 1) unintended changes in non-target areas when regenerating the entire image; 2) guidance solely by a reference image or text descriptions; and 3) time-consuming fine-tuning, which limits their practical application. In response, we introduce a tuning-free framework for simultaneous text-image-guided image customization, enabling precise editing of specific image regions within seconds. Our approach preserves the semantic features of the reference image subject while allowing modification of detailed attributes based on text descriptions. To achieve this, we propose an innovative attention blending strategy that blends self-attention features in the UNet decoder during the denoising process. To our knowledge, this is the first tuning-free method that concurrently utilizes text and image guidance for image customization in specific regions. Our approach outperforms previous methods in both human and quantitative evaluations, providing an efficient solution for various practical applications, such as image synthesis, design, and creative photography.
The same record, over MCP at https://syntology.ai/mcp:
get_harvested_code_for_paper("2403.12658")
get_code_for_paper("2403.12658")
have("2403.12658")
Connect an agent — have() is free.