mirror of
https://github.com/vladmandic/automatic
synced 2026-09-19 09:14:35 +02:00
+8
-3
@@ -2,13 +2,14 @@
|
||||
|
||||
## TODO
|
||||
|
||||
- video API: video, text2image, image2image
|
||||
- MiniMax-H3: geo-locked
|
||||
- LTX-2.5
|
||||
- Group offloading in 16gb
|
||||
|
||||
## Update for 2026-08-13
|
||||
|
||||
- **Models**
|
||||
- [MiniMax H3](https://huggingface.co/MiniMaxAI/MiniMax-H3) in *base* and *ref* variants
|
||||
MiniMax-H3 is an amazing, but absolutely massive at 32B text-encoder and 33B transformer video model
|
||||
for details, see [MiniMax wiki page](wiki/MiniMax)
|
||||
- **Detailer**: Pretty much *detailer.next* :)
|
||||
Detailer detection models were traditionally *YOLO* models, but now we can also use:
|
||||
- [Facebook-SAM3](https://huggingface.co/facebook/sam3) hybrid promptable concept segmentation and detection network
|
||||
@@ -31,6 +32,10 @@
|
||||
- log long torch autotune operations
|
||||
- utilize `torch.accelerator` where available
|
||||
- add `SD_DIFFUSERS_DEBUG` and `SD_TRANSFORMERS_DEBUG` env variables to trace diffusers and transformers internal operations
|
||||
- **API**
|
||||
- full support for video generation using api
|
||||
*note*: video api uses async workflow where you submit request and then later download the result
|
||||
new endpoints: `/sdapi/v1/video`, `/sdapi/v1/video/models`, `/sdapi/v1/video/file`
|
||||
- **Removed**
|
||||
- remove DirectML support
|
||||
latest release was over 2 years ago and is not compatible with modern frameworks
|
||||
|
||||
@@ -664,9 +664,6 @@ class ScriptRunner:
|
||||
s.report()
|
||||
|
||||
def process(self, p: StableDiffusionProcessing, **kwargs):
|
||||
from modules import shared
|
||||
if not shared.sd_loaded:
|
||||
return None
|
||||
s = ScriptSummary('process')
|
||||
for script in self.alwayson_scripts:
|
||||
try:
|
||||
@@ -679,9 +676,6 @@ class ScriptRunner:
|
||||
s.report()
|
||||
|
||||
def process_images(self, p: StableDiffusionProcessing, **kwargs):
|
||||
from modules import shared
|
||||
if not shared.sd_loaded:
|
||||
return None
|
||||
s = ScriptSummary('process_images')
|
||||
processed = None
|
||||
for script in self.alwayson_scripts:
|
||||
@@ -710,9 +704,6 @@ class ScriptRunner:
|
||||
s.report()
|
||||
|
||||
def process_batch(self, p: StableDiffusionProcessing, **kwargs):
|
||||
from modules import shared
|
||||
if not shared.sd_loaded:
|
||||
return None
|
||||
s = ScriptSummary('process-batch')
|
||||
for script in self.alwayson_scripts:
|
||||
try:
|
||||
|
||||
Reference in New Issue
Block a user