Use MOSS-Image with the smaller Qwen3.5-4B text encoder. This lightweight adapter projects 4B text features into the original 9B conditioning space, helping reduce prompt-encoding memory requirements. Fitted on 1,600 prompts. Highly experimental.<h3 id="requirements">Requirements</h3><ul><li><a target="_blank" rel="ugc" href="https://github.com/bakapotatolord/moss-image-smaller-encoder">MOSS Smaller Encoder ComfyUI node</a></li><li><a target="_blank" rel="ugc" href="https://huggingface.co/beycanai/Qwen-LM/blob/main/Qwen3.5-4B_int8_convrot.safetensors">Qwen3.5-4B INT8 ConvRot text encoder</a></li><li>Optional: <a target="_blank" rel="ugc" href="https://huggingface.co/lodestones/Kromoss/blob/main/8-steps-dmd2-lora.safetensors">Lodestones’ 8-step DMD2 LoRA</a> for faster generation. Place it in <code>ComfyUI/models/loras</code> and load through Load LoRA.</li></ul><h3 id="instructions">Instructions</h3><ol><li>Clone the custom node into <code>ComfyUI/custom_nodes</code>.</li><li>Download this adapter and place it in <code>ComfyUI/models/moss_encoder_adapters</code> (create the folder if missing).</li><li>Place the Qwen3.5-4B encoder in <code>ComfyUI/models/text_encoders</code>.</li><li>Restart ComfyUI.</li><li>Select the 4B encoder in Load CLIP, with type set to moss_image.</li><li>Connect Load CLIP → MOSS Smaller Encoder Adapter → CLIP Text Encode, selecting this adapter in the adapter node (workflow available in images as well)</li></ol>This is a projection adapter, not a LoRA. Results will differ from the original 9B encoder.

MOSS-Image 4B Text Encoder Adapter (9B -> 4B)

mossImage4BTextEncoder_1600Fit.safetensors

1600 fit

Use MOSS-Image with the smaller Qwen3.5-4B text encoder. This lightweight adapter projects 4B text features into the original 9B conditioning space, helping reduce prompt-encoding memory requirements. Fitted on 1,600 prompts. Highly experimental.<h3 id="requirements">Requirements</h3><ul><li><a target="_blank" rel="ugc" href="https://github.com/bakapotatolord/moss-image-smaller-encoder">MOSS Smaller Encoder ComfyUI node</a></li><li><a target="_blank" rel="ugc" href="https://huggingface.co/beycanai/Qwen-LM/blob/main/Qwen3.5-4B_int8_convrot.safetensors">Qwen3.5-4B INT8 ConvRot text encoder</a></li><li>Optional: <a target="_blank" rel="ugc" href="https://huggingface.co/lodestones/Kromoss/blob/main/8-steps-dmd2-lora.safetensors">Lodestones’ 8-step DMD2 LoRA</a> for faster generation. Place it in <code>ComfyUI/models/loras</code> and load through Load LoRA.</li></ul><h3 id="instructions">Instructions</h3><ol><li>Clone the custom node into <code>ComfyUI/custom_nodes</code>.</li><li>Download this adapter and place it in <code>ComfyUI/models/moss_encoder_adapters</code> (create the folder if missing).</li><li>Place the Qwen3.5-4B encoder in <code>ComfyUI/models/text_encoders</code>.</li><li>Restart ComfyUI.</li><li>Select the 4B encoder in Load CLIP, with type set to moss_image.</li><li>Connect Load CLIP → MOSS Smaller Encoder Adapter → CLIP Text Encode, selecting this adapter in the adapter node (workflow available in images as well)</li></ol>This is a projection adapter, not a LoRA. Results will differ from the original 9B encoder.Reuploaded due to a mistake with the uploaded files

Files with SHA256: df5ed7fa15f4c628ad6cb3539d9cd16d68f198df9cd1e07740e9bd7b162c954b