update changelog and todo

Signed-off-by: vladmandic <mandic00@live.com>

CHANGELOG.md

@@ -1,6 +1,6 @@
 # Change Log for SD.Next
 
-## Update for 2026-04-12
+## Update for 2026-04-13
 
 - **Models**
   - [Zeta-Chroma](https://huggingface.co/lodestones/Zeta-Chroma) pixel-space diffusion transformer image model  
@@ -21,8 +21,6 @@
     integrates Qwen2.5-VL text encoder and Flow Matching scheduler for high-quality in-context edits  
     enables multimodal understanding and conditional image refinement  
     category: *base*  
-- **Caption & Prompt Enhance**
-  - [Google Gemma 4] in *E2B* and *E4B* variants as well as *heretic* fine-tune
 - **Compute**
   - **ROCm** futher work on advanced configuration and tuning, thanks @resonantsky  
     now covers both ROCm on Windows and Linux  
@@ -30,6 +28,7 @@
   - **OpenVINO**  
     update to `openvino==2026.1.0`  
     default device is now `AUTO`, override which devices are allowed to be used in *settings -> backend settings*  
+    perform detection of fp16 capabilities instead of forcing fp32 by default  
     add *experimental* support for `NPU` devices, typically available on AI-PC/Copilot-PC devices  
   - **SDNQ** add quant support to `nn.Embedding` type  
 - **Features**
@@ -40,6 +39,10 @@
   - enhanced filename pattern processing  
     allows for any *processing* property name (as defined in `modules/processing_class.py` and saved to `ui-config.json`)  
     allows for any *settings* property name (as defined in `modules/ui_definitions.py` and saved to `config.json`)  
+- **UI**
+  - `gallery` send-to button advanced options with right-click
+- **Caption & Prompt Enhance**
+  - [Google Gemma 4] in *E2B* and *E4B* variants as well as *heretic* fine-tune
 - **Agents**
   framework for AI agent based work in `.github/`  
   *note*: all skills are agent-model agnostic  

TODO.md

@@ -11,7 +11,6 @@
 - Test: Lumina-DiMOO
 - Test: Step1X-Edit
 - Code: prompt encode for Bria-FIBO: <https://github.com/Bria-AI/Fibo-Edit/blob/master/src/fibo_edit/fibo_edit_vlm.py>
-- Code: OpenVINO device selection: auto/cpu/gpu/npu
 - Port: ERNIE-Image (merged, unpublished)
 - Port: NucleusMoE-Image (merged, unpublished)
 - Port: JoyAI-Image-Edit (in-progress, published)
@@ -41,6 +40,7 @@
 - Deploy: Lite vs Expert mode
 - Engine: [mmgp](https://github.com/deepbeepmeep/mmgp)
 - Engine: `TensorRT` acceleration
+- Engine: [DiffSynth-Engine](https://github.com/modelscope/DiffSynth-Engine)
 - Feature: Auto handle scheduler `prediction_type`
 - Feature: Cache models in memory
 - Feature: JSON image metadata
@@ -77,8 +77,8 @@ TODO: Investigate which models are diffusers-compatible and prioritize!
 
 ### Image
 
+- [UltraFlux](https://huggingface.co/Owen777/UltraFlux-v1)
 - [Mugen](https://huggingface.co/CabalResearch/Mugen)
-- [NucleusMoe](https://github.com/huggingface/diffusers/pull/13317)
 - [Liquid](https://github.com/FoundationVision/Liquid)
 - [nVidia Cosmos-Predict-2.5](https://huggingface.co/nvidia/Cosmos-Predict2.5-2B)
 - [Liquid (unified multimodal generator)](https://github.com/FoundationVision/Liquid)
@@ -111,9 +111,17 @@ TODO: Investigate which models are diffusers-compatible and prioritize!
 - [LivePortrait](https://github.com/KwaiVGI/LivePortrait): Efficient portrait animation system with high stitching and retargeting control
 - [Magi (SandAI)](https://github.com/SandAI-org/MAGI-1): High-quality autoregressive video generation framework
 - [Ming (inclusionAI)](https://github.com/inclusionAI/Ming): Unified multimodal model for processing text, audio, image, and video
+- [HummingbirdXT](https://huggingface.co/amd/HummingbirdXT)
 
 ### Other/Unsorted
 
+- [RamTorch](https://github.com/lodestone-rock/ramtorch)
+- [FaceFusion](https://github.com/facefusion/facefusion)
+- [FaceClip](https://huggingface.co/ByteDance/FaceCLIP)
+- [TwinFlow](https://github.com/inclusionAI/TwinFlow)
+- [RegionE](https://github.com/Peyton-Chen/RegionE)
+- [T5Gemma Adapter](https://huggingface.co/Minthy/Rouwei-T5Gemma-adapter_v0.2)
+- [FreeFuse](https://github.com/yaoliliu/FreeFuse)
 - [OneReward](https://github.com/bytedance/OneReward)
 - [ByteDance DreamO](https://huggingface.co/ByteDance/DreamO)
 - [DiffusionForcing](https://github.com/kwsong0113/diffusion-forcing-transformer): Full-sequence diffusion with autoregressive next-token prediction