logo
Start creating

How to run catalog reshoots at scale with Picsart Skills

SKILLS5 minAdvanced

Generate thousands of product variants in batch mode with parallel processing and automatic resume.

How to run catalog reshoots at scale with Picsart Skills

What you'll learn

  • How to set up batch generation for thousands of SKUs
  • How to run parallel processing to speed up large jobs
  • How to resume failed batches without starting over
  • How to sync generated images directly to cloud storage

What is a catalog reshoot?

A catalog reshoot generates new lifestyle or product photography for your entire inventory without physically photographing anything. You feed the Skill your existing product images and a list of desired settings (backgrounds, lighting, angles, contexts), and it creates thousands of variants in batch mode. Parallel processing handles scale, automatic resume handles failures, and cloud sync delivers files straight to your asset library. Think of it like hiring a photo studio that works 24/7 and never needs coffee breaks.

Common use cases

  • E-commerce: Generate lifestyle variants for every SKU (in-context, on-model, seasonal)
  • Fashion brands: Create lookbook images without physical photoshoots
  • Seasonal campaigns: Regenerate catalog with holiday or themed backgrounds
  • A/B testing: Generate multiple variants per product to test conversion rates
  • Marketplace sellers: Create professional product photos from basic snapshots
  • Print catalogs: Batch-generate high-res images for quarterly catalog updates

Run your catalog reshoot step by step

STEP 1: Set up your batch manifest

  • On web: Go to picsart.com/cli → Install Skills → Create a JSON manifest file
  • On mobile: Not available — Skills runs in server or CI environments for batch jobs
Get Picsart Skills

STEP 2: Configure processing and output settings

Define how the batch job runs:

  • Concurrency level: Number of parallel generations (higher = faster, more resource-intensive)
  • Variant count: How many variations to generate per SKU (e.g., 5 backgrounds × 3 angles = 15 images)
  • Output format: File type, resolution, and naming convention
  • Cloud sync: Auto-upload to S3, Google Cloud Storage, or your CDN
  • Resume mode: Enable auto-resume for failed items without reprocessing completed ones

STEP 3: Run the batch job

Execute the Skill with your manifest. Progress updates show in real-time, tracking completed items, active generations, and any failures. If the process stops (network issue, rate limit, server restart), resume mode picks up where it left off without regenerating finished images. Completed files sync to your cloud storage as they finish, so you can start using them before the entire job completes.

STEP 4: Review quality and deploy

Check a sample of generated images across different SKUs and variants: Not satisfied with specific variants? Filter your manifest to just the problem items, adjust prompts or settings, and regenerate only those. The rest of your catalog stays untouched.

  • Verify product details are preserved (colors, textures, logos, labels)
  • Check that backgrounds and contexts match your brand aesthetic
  • Confirm file naming and folder structure match your asset management system
Start batch reshoot

Tips for best results

💡 Start with a test batch

Run 50-100 SKUs as a test before processing thousands. This lets you catch style inconsistencies, naming errors, or output issues early. Fixing a batch of 50 is annoying. Fixing 5,000 is a disaster.

💡 Tune concurrency based on your infrastructure

Higher concurrency speeds up processing but uses more memory and network bandwidth. Start conservative (5-10 parallel jobs) and increase if your system handles it without timeouts or rate limits.

💡 Use variant templates for consistency

Define a set of standard variant templates (white background, lifestyle, on-model, seasonal) that apply across all products. This ensures catalog-wide consistency and makes it easier to add new SKUs later.

💡 Monitor credit usage during long jobs

Large batches can consume thousands of credits. Enable spend alerts or dry-run mode to estimate costs before execution. The Skill shows credit projections based on your manifest, so you know total cost upfront.

Batch processing settings

  • Low concurrency (1-5): Best for small batches or limited server resources. Slowest but most stable.
  • Medium concurrency (6-15): Good balance for most catalog reshoots. Handles 1000+ SKUs in reasonable time.
  • High concurrency (16-30): Fast processing for large catalogs. Requires robust infrastructure and higher rate limits.
  • Resume on failure: Automatically skips completed items if batch job stops. Essential for large reshoots.
  • Cloud sync: Auto-uploads finished images to S3, GCS, or CDN. Eliminates manual file transfer.
  • Dry run mode: Validates manifest and estimates credit cost without generating images. Use before big jobs.

Frequently asked questions

With medium concurrency (10 parallel jobs), generating 1000 SKUs × 5 variants each takes around 3-5 hours. Higher concurrency cuts time significantly, but requires more server resources. Resume mode means you can pause and continue later without losing progress.

Resume mode automatically skips completed items when you restart the job. Only failed or unprocessed SKUs regenerate. You don't lose progress or waste credits re-generating images that already finished. Progress logs show exactly where the job stopped.

Yes. Your manifest can specify variant templates per SKU or category. Apparel gets on-model variants, furniture gets room-context variants, and accessories get close-up detail shots. Each product type can have its own generation rules within the same batch.

Run the Skill in dry-run mode with your manifest. It calculates total credit cost based on SKU count, variant count, and model settings. You see the exact number before committing. A typical reshoot costs 2-4 credits per generated image.

Ready to reshoot your catalog?

Generate thousands of product variants in parallel with automatic resume and cloud sync.

Start reshoot