<h1 id="flashvsr-video-upscaling-workflow">FlashVSR Video Upscaling Workflow</h1>A ComfyUI workflow for video frame interpolation + AI video super-resolution, mainly tested on anime / illustration-style videos.<h2 id="recommended-workflow">⭐ Recommended Workflow</h2>For low-FPS videos, I recommend:Input Video → Frame Interpolation ×2 → FlashVSR → Video CombineFor example:<code>16 FPS → 32 FPS → FlashVSR 4× → 32 FPS</code><h3 id="interpolate-before-flashvsr">⚠️ Interpolate BEFORE FlashVSR</h3>After testing both processing orders, I found that interpolating the video before FlashVSR significantly reduces visible motion ghosting in low-FPS sources.❌ FlashVSR → VFI<ul><li>Faster at 4K</li><li>More noticeable ghosting/trailing around fast-moving objects</li><li>Particularly noticeable around hands and arms</li></ul>✅ VFI → FlashVSR<ul><li>Much less noticeable ghosting in my testing</li><li>Better temporal consistency</li><li>Recommended for low-FPS sources</li><li>Significantly more expensive at 4K</li></ul><blockquote>Important: This does not completely eliminate ghosting. FlashVSR itself can still produce subtle temporal artifacts in difficult high-motion scenes. The recommended order simply makes them much less noticeable in my testing.</blockquote><hr /><h2 id="performance-settings">⚡ Performance Settings</h2>After testing the Advanced FlashVSR settings on an RTX 5080 16GB, I found a much faster configuration than the default settings.<h3 id="recommended-16gb-vram-settings">Recommended 16GB VRAM settings</h3><pre><code></code></pre><pre><code>tile_size = 224
tile_overlap = 16
sparse_ratio = 1.8
kv_ratio = 2.2</code></pre>Other settings can remain at their defaults.With these settings, my ~3-second 16 FPS source took approximately 8 minutes for 4× 4K upscaling using:<code>VFI ×2 → FlashVSR</code>This is a personal benchmark on an RTX 5080 16GB, so actual performance will vary depending on GPU, VRAM, source resolution, video length and other settings.<h3 id="vram-tile-size">⚠️ VRAM / Tile Size</h3>Larger <code>tile_size</code> does not necessarily mean faster processing.In my testing:<ul><li><code>224</code> → ~73% VRAM usage and very good performance</li><li><code>256</code> → ~95–100% VRAM usage, but little/no speed improvement</li><li><code>288</code> → caused severe VRAM pressure and was significantly slower</li></ul>Once VRAM is exhausted, system RAM starts being used and performance can drop dramatically.For 16GB GPUs, <code>224</code> is currently my recommended starting point.<hr /><h2 id="color-fix">🎨 Color Fix</h2>I recommend keeping <code>color_fix</code> enabled.Without color correction, I observed noticeable color shifts in dark scenes, including changes to the appearance of the night sky/background.<hr /><h2 id="about-ghosting">👻 About Ghosting</h2>Some subtle ghosting may still remain, especially during very fast motion.This appears to be a limitation of the current FlashVSR processing rather than something that can be completely solved by changing the VFI model.The best result I found was:VFI ×2 → FlashVSRrather than:FlashVSR → VFIThe latter is faster at 4K, but produces more obvious motion artifacts in my testing.<hr /><h2 id="my-test">🧪 My Test</h2>Source video:<ul><li>Anime / illustration style</li><li>16 FPS</li><li>~3 seconds</li></ul>Recommended pipeline:<pre><code></code></pre><pre><code>Load Video
 ↓
VFI ×2
 ↓
FlashVSR 4×
 ↓
Video Combine</code></pre>Tested on:RTX 5080 16GBOptimized 4K processing time:~8 minutesResults:<ul><li>✅ No obvious visual artifacts in normal viewing</li><li>✅ Much less noticeable motion ghosting</li><li>✅ Good 4K detail</li><li>✅ Significantly faster than the default Advanced settings in my testing</li><li>⚠️ Subtle ghosting can still be visible when examining individual frames closely</li></ul><hr /><h2 id="before-you-render">⚠️ Before You Render</h2>Please test a short clip first.4K video processing can become extremely expensive, particularly when frame interpolation is performed before super-resolution.The optimal settings depend on:<ul><li>Source FPS</li><li>Source resolution</li><li>Motion speed</li><li>Scene complexity</li><li>GPU / VRAM</li><li>VFI model</li></ul>For GPUs with limited VRAM, 2K is recommended as a starting point.<hr /><h3 id="requirements">Requirements</h3><ul><li>ComfyUI</li><li>ComfyUI-Frame-Interpolation</li><li>FlashVSR v1.1</li><li>A compatible VFI model (ATM-VFI, RIFE, GMFSS Fortuna, etc.)</li><li>NVIDIA GPU</li></ul>GMFSS Fortuna requires additional dependencies such as CuPy.

Files with SHA256: 5e9baeed3a2a75cb1e1344266c05b5cc45612f76cf91c8c953489231c1506a88