Commit Graph
48 Commits
Author SHA1 Message Date
blepping 5ea059bed5 Bug fixes
More expression tensor operations
Make the return expression handler actually work
2026-07-11 10:44:15 -06:00
blepping ee59df94e3 New Perlin, expression filtered patches, expression methods 2026-06-23 11:15:31 -06:00
blepping d8574bae2d Mega update (#8)
* Weird experiments with sampler blending

* Mega sync

* Use ComfyUI union types for wildcard inputs when available

* Add weoon wavelet sampler

* Fix noise sampler caching not considering immiscible settings
Allow specifying alt custom noise/immiscible settings for samplers that internally add noise
AFS support at the merge sampler level
Allow the Weoon internal step to use ETA
Add t_copysign expression function

* Start updating documentation

* Add ImmiscibleReference node, other changes

* Respect factor in ImmiscibleReference noise

* Fixing normalizing in ImmiscibleReference noise

* Handle case where there are no RGB factors (i.e. audio models)

* Fix rectified flow type detection for non-Flux models

* Add more tensor ops, add OCS ApplyExpressionImage/Latent nodes

* Rename ApplyExpression nodes to ApplyFilter, expression QoL improvements

* Fix apply filter nodes definitions

* Make t_noise handler work for image-latents.

* Add t_new_like expression function

* Make it possible to comment out lines with hash in expressions

* Add gradient estimation, pingpong and res_multistep samplers.
Add pingpong group merge method.
Remove model call caching stuff.
Allow defining pre_cfg and post_cfg filters.
Handle filters changing the latent shape better.
Allow disabling detecting out of order sigmas as restart sampling.

* Sync current updates (which I am too lazy to describe individually)

* Add ExpressionFilteredNoise node
Improvements to immiscible noise handling/expanded features
2025-08-09 07:28:10 -06:00
blepping 716b0d749a Improve compatibility with older Python versions, part 2 2024-09-14 18:50:49 -06:00
blepping 03fe373c10 Improve compatibility with older Python versions 2024-09-14 09:09:58 -06:00
blepping 2e4bf47017 Fix cfgpp and alt_cfgpp_scale for Euler 2024-09-12 02:06:23 -06:00
blepping 77747875b0 Add distance, rkf45 and rk_dynamic samplers
Fix issue with samplers noising a second internal step (only dpmpp_sde was affected)

Pass the sigmas to the noise sampler when possible, may make brownian noise work better

Add eta_retry_increment OCS Substeps parameter, when set, if the ETA calculation NANs it will try with a lower value

Fix an issue with the lookahead substep merge method eating its settings

Internal cleanups and refactoring
2024-09-10 15:05:35 -06:00
blepping cb773cd851 Allow overriding CFG scale in parameters
Try to make sure uncond gets generated when needed
2024-09-07 06:35:52 -06:00
blepping 67a479a65d Add dynamic step and group methods 2024-09-05 09:27:45 -06:00
blepping 2528e6ae79 Allow renaming parameter inputs, document OCS Param and OCS MultiParam nodes 2024-09-05 06:32:38 -06:00
blepping 4ed521fdcc Add some information for Flux users to README 2024-09-03 18:41:35 -06:00
blepping 0cc120c0ab Fix s_noise in lookahead merge method block (still does not work with RF) 2024-09-03 18:36:57 -06:00
blepping d65c096438 Add lookahead substep merge method
Rename normal substep merge method to supreme_avg
2024-09-03 17:32:12 -06:00
blepping cc30fde46c Initial implementation of ancestral sampling for rectified flow models, supports most samplers 2024-09-01 17:24:56 -06:00
blepping 1618093e22 Do not require eta/s_noise set in main sampler for no reason 2024-09-01 17:01:04 -06:00
blepping d5031a8ea7 Add setting to allow skipping uncond on models like Flux 2024-09-01 13:27:29 -06:00
blepping ec0ad9faf3 Add noisy and diff preview_mode options 2024-08-30 17:42:27 -06:00
blepping 76091dc006 Un-derp initial noise cache settings, for old behavior set caching: false in noise block 2024-08-30 14:01:14 -06:00
blepping a734aeb5a2 Make pre-restart refs available to expressions in restart filters 2024-08-30 13:49:40 -06:00
zer0int e58189e5f0 Fix generator expression error w/ list comprehension (#7)
* Fix incompatibility with older Python versions in expression flip function handler
2024-08-27 14:43:48 -06:00
blepping f1a8decc55 Fix != expression operator 2024-08-27 10:29:20 -06:00
blepping 74421f7fbc Perlin 2D/3D and other features update (#6)
* Perlin3D and 2D implementation

* Fix Euler not allowing `alt_cfgpp_scale`

* Enable alt_cfgpp_scale for `dpmpp_2s`, `dpmpp_sde` and RES samplers
2024-08-27 08:23:32 -06:00
blepping bc1e4b0874 Fix setting group parameters 2024-08-17 14:45:39 -06:00
blepping 3e73d2e352 Fix uncond preview_mode 2024-08-17 08:20:55 -06:00
blepping 0aa45f98ff Allow setting a preview_mode in groups 2024-08-17 07:45:19 -06:00
blepping aac7a05bb2 Fix issue where batches didn't work with Diffrax solver
Fix some other issues related to ODE/SDE solvers and batches
2024-08-17 07:23:51 -06:00
blepping ac1951101f Add TAESD encode/decode and some image handling function for expressions
Documentation updates

Allow filtering model input result
2024-08-16 11:30:04 -06:00
blepping fdbbd970e7 Adjust noise scaling approach 2024-08-16 11:09:54 -06:00
blepping 9db4fbe514 Fix noise sampler reset for really reals this time 2024-08-16 11:09:19 -06:00
blepping 2d8125e011 Allow dash in expression symbols 2024-08-16 11:07:58 -06:00
blepping 076d80918a Add tooltip metadata to nodes 2024-08-15 08:40:15 -06:00
blepping bf43d36bac Allow using nnlatentupscale in expressions if available 2024-08-15 07:53:46 -06:00
blepping fd5d0ef6a1 Fix noise cache update to work better when the latent size changes 2024-08-15 05:01:13 -06:00
blepping 0595a6d304 Fix another fstr in fstr issue 2024-08-09 08:52:51 -06:00
blepping 305720c66e Fix issue with fstr in fstr 2024-08-09 08:33:51 -06:00
blepping f0e52ec054 Make restarting work better when the latent changes sizes during sampling 2024-08-08 19:35:51 -06:00
blepping 442bde294a Add t_shape expression function
Fix width/height order for t_scale in absolute mode
2024-08-08 14:10:25 -06:00
blepping 764f04a6a7 Fix t_noise expression function 2024-08-08 07:52:40 -06:00
blepping 2481e7263b Allow setting temporary variables in expressions
Improve sequence operator in expressions

Documentation improvements

Turn down eval debug spam a bit

Add ternary operator

Support static eval for constants in some more operator types
2024-08-08 06:27:53 -06:00
blepping 470b38231f Refactor (#1)
Refactor all the things!
2024-08-04 11:28:06 -06:00
blepping 8d0716623c Fix substep debug print for average and sample merge types 2024-06-01 14:54:01 -06:00
blepping 81f0de654e Fix sample merge strategy 2024-05-31 14:01:39 -06:00
blepping b5a4bf9895 Rewrite model call caching 2024-05-31 13:21:13 -06:00
blepping bf8d525db2 Allow setting dynamic ETA and RETA, other cleanups 2024-05-30 22:51:09 -06:00
blepping be5e5d8933 All sorts of fun stuff 2024-05-30 19:47:44 -06:00
blepping 05dcec2985 Fix incorrectly caching first model eval 2024-05-30 08:14:55 -06:00
blepping aa3e12ae34 Initial implementation 2024-05-30 07:47:43 -06:00
blepping 360ba4842a Initial commit 2024-05-30 02:48:23 -06:00