Troubleshooting
Most validation messages stop a render before allocating GPU memory or starting a long video encode. Locate the message below to fix the exact pipeline, codec, or hardware constraint.
Stage Ordering & SDR / HDR Errors
Section titled “Stage Ordering & SDR / HDR Errors”| Validation / Error Message | Cause | Fix |
|---|---|---|
| “Denoising requires SDR input. Place it before HDR conversion.” | Denoising (ClearMagnetar), Deblurring (FocusPulsar), Upscale (PeakMagnetar / OptiPulsar / VisionNext / RazorNext / DLSS / RTX), Scaling, or Sharpening appears below RTX Video HDR—or your video input is already HDR. |
Drag the affected card above RTX Video HDR in the Pipeline rail. For native HDR source videos, disable SDR-only cleanup/upscaling cards or use DLSS Neural Rendering, DLSS Ray Reconstruction, DLSS Frame Generation / FSR Frame Generation, Coloring, or ReShade (which support HDR). |
| “Use one frame-generation engine per workflow. Select DLSS, AMD FSR or VFX Frame Generation.” | Multiple frame-generation stages are active simultaneously in the pipeline. | Keep only one frame-generation engine enabled (for example, either DLSS Frame Generation or AMD FSR 3, not both). |
| “Frame Generation target FPS must exceed the frame rate entering this card. Choose a higher target FPS.” | Target FPS is less than or equal to the clip’s frame rate entering the card. | Switch Frame Rate Mode to Multiplier (2x–8x), or pick a higher Target FPS value. |
| “RTX Video HDR requires SDR at this position.” | The input video is already HDR (PQ/HLG), or HDR conversion appears twice. |
Disable RTX Video HDR on native HDR videos—your source is already in HDR color space. |
| “Ray Reconstruction requires explicit PQ or HLG signaling for HDR video.” | An HDR video lacks smpte2084 (PQ) or arib-std-b67 (HLG) transfer metadata and RTX Video HDR is off. |
Re-mux/tag transfer characteristics cleanly, or process SDR footage instead. |
| “HDR output requires 10-bit HDR Mode in Export Settings.” / “HDR requires a 10-bit or higher video codec.” | RTX Video HDR or an HDR source video is exporting with Preserve / Export 10-bit HDR unchecked or with an 8-bit codec (H.264, software H.265 Kvazaar). |
Check Preserve / Export 10-bit HDR in Export Settings and switch Video Codec to H.265 (NVIDIA NVENC), AV1, AV1 (NVIDIA NVENC), VP9, ProRes Proxy/HQ, or FFV1 Lossless RGB 10-bit. |
Resolution & Dimension Errors
Section titled “Resolution & Dimension Errors”| Error / Symptom | Cause | Fix |
|---|---|---|
| “Software HEVC requires output dimensions divisible by 8.” | Selected software H.265 (libkvazaar), which requires both output width and height to be multiples of 8. |
Choose H.265 (NVIDIA NVENC), AV1, VP9, ProRes, or FFV1, or resize your output so both axes divide cleanly by 8. |
| “Video export requires even output dimensions.” / “This video codec requires even output dimensions.” | H.264, H.265, or AV1 output has an odd pixel width or height. |
Switch to VP9, ProRes, or FFV1 (which accept odd dimensions), or set an even output size in Scaling / Upscale. |
| “DLSS output exceeds the 16384-pixel texture limit.” / “FSR Frame Generation supports up to 16384 pixels per side.” | A frame entering DLSS NR, DLSS Frame Generation, or AMD FSR 3 exceeds 16,384 px on width or height (or is below 64 px). |
Reduce your upscaling factor before the temporal/DLSS card, or place Scaling / Upscale after DLSS / Frame Generation (up to the 32,768 px pipeline ceiling). |
| “WebP format dimensions … must be positive and at most 16383 pixels per side.” | Exporting an image wider or taller than 16,383 px as WebP. |
Export as PNG, TIFF, JPEG XL, or JPEG, or downscale slightly under 16,383 px. |
“…cannot encode WxH; its maximum dimensions are …” (4096x4096 / 8192x8192) |
H.264 exceeds 4096 × 4096 (or 36,864 macroblocks), or NVENC HEVC/AV1 exceeds 8192 × 8,192. |
Switch to software AV1 (16,384 × 8,704), VP9, ProRes, or FFV1 for >8K video exports. |
GPU, HAGS, Drivers & NeuroFrame Runtime
Section titled “GPU, HAGS, Drivers & NeuroFrame Runtime”- DLSS / RTX buttons disabled or
"Requires an NVIDIA RTX GPU with compute capability 7.5 or newer": Verify an RTX 20 series or newer GPU is selected under Settings > Hardware & GPU Acceleration > AI Processing GPU (Tensor Cores / NGX) (or--gpu), and update to a current NVIDIA Game Ready or Studio driver. - DLSS Frame Generation stutters, fails to initialize, or reports unsupported hardware: Enable Hardware-Accelerated GPU Scheduling (HAGS) in Windows Graphics settings and reboot Windows.
- FSR 3 Frame Generation unavailable: Check Settings > AI Processing GPU (DirectX 12 / FidelityFX) or run
VE_CLI.exe devicesto verify your DirectX 12 DXGI adapter reportsFSR 3: available. On laptops with integrated + discrete GPUs, pick the high-performance adapter explicitly rather thanauto. - ClearMagnetar / FocusPulsar / VisionNext / RazorNext / PeakMagnetar / OptiPulsar missing runtime: Open Models in the header bar and install both the model package and the NeuroFrame Runtime matching your GPU (
CUDA,XPU, orROCm / HIP).
ReShade Previews, Depth & Temporal Warmup
Section titled “ReShade Previews, Depth & Temporal Warmup”- Missing
.fxshaders or textures: Ensure both Shader Search Folder (.fx / .fxh) and Texture & LUT Search Folder (separated by;) include the rootShadersandTexturesdirectories referenced by your.inipreset. - First few frames of a scrubbed video preview differ slightly from full render: When previewing mid-clip with temporal stages (ReShade Temporal Stability, DLSS Ray Reconstruction, FSR 3 Frame Generation, or RazorNext video), Visual Enhancer renders a short hidden temporal warmup context preceding your seek timestamp and discards it before displaying the clip—matching final export temporal convergence without leaving startup artifacts.
- Depth-based effects (
qUINT_mxao,DisplayDepth, etc.) appear flat: Verify Depth Input is set to Auto or Always Generate AI Depth, and ensure theMoGe-2model (moge-2-vitb) is downloaded in Models.
Disk Space & Cache Cleanup
Section titled “Disk Space & Cache Cleanup”- Clean Cache button is disabled: Wait for active previews, renders, or LUT analyses to finish (
Clean Cacheautomatically unlocks when the pipeline is idle and protects files currently displayed in viewports or queues). - Inspecting failures: Click Settings > Open Logs (or check
<app-folder>\logs\) to inspect detailed FFmpeg, CUDA, DirectX 12, and Python worker diagnostics for any failed job.