Cdf anomalygen vs Physical AI defect image generation
Cdf anomalygen is an open-source agent skill for design and media. Physical AI defect image generation is an open-source agent skill for design and media.
- Cdf anomalygen is best for run the anomalygen pipeline on a phone screen defect dataset.
- Physical AI defect image generation is best for generate synthetic solder bridge or scratch defect images for a pcb inspection dataset.
- What it is
- Cdf anomalygen is an open-source agent skill for design and media.Physical AI defect image generation is an open-source agent skill for design and media.
- Best for
- Run the AnomalyGen pipeline on a phone screen defect datasetGenerate synthetic solder bridge or scratch defect images for a PCB inspection dataset
- Kind
- SkillSkill
- Pricing
- Open sourceOpen source
- Platforms
- Claude Code, Codex, Cursor, Gemini CLI, GitHub Copilot, Windsurf, GooseClaude Code, Codex, Cursor, Gemini CLI, GitHub Copilot, Windsurf, Goose
- GitHub stars
- 4K4K
- Last commit
- Nov 2026Nov 2026
- License
- Apache-2.0Apache-2.0
- Open source
- Good for
- Run the AnomalyGen pipeline on a phone screen defect dataset; Fine-tune an anomaly-generation model for PCB defects; Generate 25 synthetic defect images from an existing checkpointGenerate synthetic solder bridge or scratch defect images for a PCB inspection dataset; Label defects on new real AOI photos of a supported circuit board; Finetune an AnomalyGen checkpoint for metal surface defects
What Cdf anomalygen does
Dataset layout validation
Checks that clean images, defect images and masks follow the expected folder contract before any stage runs.
Mask placement backends
Places defect masks either through a vision-language model agent loop or a classic allocation and pairing pipeline.
Model fine-tuning
Runs a detached fine-tune from a recipe, with checkpoint selection from validation KPIs.
Synthetic defect generation
Produces defect images from a checkpoint and testcase file, with guardrail-blocked samples logged separately.
KPI evaluation
Scores generated batches against real data using nearest-neighbour and related metrics.
Quality refinement and labeling
Runs search rounds to refine outputs, then exports COCO annotations and classification layouts.
What Physical AI defect image generation does
Day 0 cold-start generation
Builds initial PCBA datasets from a CAD scene USD through ROI rendering, image-edit augmentation and AnomalyGen.
Day 1 inference and labeling
Runs defect inference and labeling on real images, with real-photo alignment for PCBA or manual ROI for metal and glass.
Finetuning workflow
Produces AnomalyGen finetuning checkpoints from raw labeled data for later Day 0 or Day 1 runs.
OSMO workflow orchestration
Provides canonical OSMO workflow YAMLs that chain each pipeline step non-interactively.
Preflight validation
Checks credentials, pod template and required storage artifacts before each submission.
Output sizing guidance
Explains how patch and crop limits map to final image counts so output volume can be planned.
Cdf anomalygen vs Physical AI defect image generation: questions
Is Cdf anomalygen or Physical AI defect image generation cheaper?
Both are open source. Compare current plans on each official site.
What is the difference between Cdf anomalygen and Physical AI defect image generation?
Cdf anomalygen is an open-source agent skill for design and media. Physical AI defect image generation is an open-source agent skill for design and media.
