<h1 id="lora-iterator-comfyui-custom-node">LoRA Iterator — ComfyUI Custom Node</h1><p><a target="_blank" rel="ugc" href="comfy-lora-iterator">https://www.theoath.studio/projects/comfy-lora-iterator?utm_source=civitai</a></p><p></p><p>A ComfyUI node that automatically steps through every LoRA in a directory or subdirectory on each generation run, allowing full LoRA testing without modifying the graph between runs.</p><p></p><hr /><h1 id="install">Install</h1><ol><li><p>Clone or copy this folder into your <code>ComfyUI/custom_nodes/</code> directory:</p></li></ol><pre><code>cd ComfyUI/custom_nodes/
git clone git@github.com:OATH-Studio/comfy-LoRA-iterator.git</code></pre><ol><li><p>Restart ComfyUI.</p></li></ol><p>The node appears under:</p><pre><code>loaders/lora → LoRA Iterator</code></pre><hr /><h2 id="inputs">Inputs</h2><p>InputTypeDescriptionmodelMODELBase model passthroughclipCLIPBase CLIP passthroughdirectoryDropdown<code>[All]</code> or a specific subdirectory inside the loras folderlora_nameDropdownCurrently selected LoRA (relative path)strength_modelFloat sliderLoRA strength applied to the diffusion modelstrength_clipFloat sliderLoRA strength applied to the CLIP encodercontrol_after_generateDropdownDefines how LoRA selection advances after each run</p><hr /><h2 id="outputs">Outputs</h2><p>OutputTypeDescriptionmodelMODELModel with LoRA appliedclipCLIPCLIP with LoRA appliedlora_nameSTRINGName of the LoRA used in the current runcurrent_indexINTZero-based index of current LoRAtotal_lorasINTTotal number of LoRAs in selected directory</p><hr /><h2 id="control_after_generate">control_after_generate</h2><p>Defines how the node advances its internal LoRA selection after each generation.</p><p>ModeBehaviourfixedAlways uses the same LoRAincrementMoves forward one LoRA per run, wraps to 0decrementMoves backward one LoRA per run, wraps to lastrandomizeSelects a random LoRA each run</p><p>Example:</p><p>You have 8 LoRAs in a directory. Set:</p><ul><li><p><code>directory = zimage</code></p></li><li><p><code>control_after_generate = increment</code></p></li></ul><p>Queue 8 runs. Each run uses the next LoRA automatically.</p><hr /><h2 id="directory-organisation">Directory Organisation</h2><p>The directory dropdown includes <code>[All]</code> and every immediate subdirectory inside:</p><pre><code>ComfyUI/models/loras/
    zimage/
        AnimeMix.safetensors
        Realism.safetensors
    sdxl/
        portrait.safetensors
    my_lora.safetensors</code></pre><ul><li><p>Selecting <code>zimage</code> limits iteration to that folder</p></li><li><p><code>[All]</code> includes every LoRA across all folders</p></li></ul><p>Prevents mixing incompatible LoRAs with base models.</p><hr /><h2 id="multiple-instances">Multiple Instances</h2><p>Each node instance tracks its own LoRA index independently using ComfyUI’s internal <code>unique_id</code>.</p><p>Multiple iterator nodes in the same graph do not interfere with each other.</p><hr /><h2 id="typical-workflow">Typical Workflow</h2><pre><code>CheckpointLoader → LoRA Iterator → KSampler → VAEDecode → SaveImage
                         ↑
            control_after_generate = increment
            queue N runs (one per LoRA)</code></pre><p>Connect <code>lora_name</code> to filename prefix logic to embed the active LoRA name into saved outputs automatically.</p>

Files with SHA256: f6d72da585e99ab343741297abec7fc8d21f0d2f2d7b8ffc144268a319cde771