100 Gemini Flash images cost 500 credits — $5.00. That is 100 unique jewelry concepts, each rendered as a photorealistic studio shot, for less than an iced coffee. The economics of batch AI generation make it practical to explore design space at a scale that was never viable with hand sketching or CAD.
What Batch Generation Costs
The credit cost scales linearly by provider. Here is the math for a 100-image batch:
| Provider | Credits per Image | 100 Images | Total Cost |
|---|---|---|---|
| Gemini Flash | 5 | 500 credits | $5.00 |
| Gemini Pro | 17 | 1,700 credits | $17.00 |
| OpenAI GPT-Image | 10 | 1,000 credits | $10.00 |
Gemini Flash is the clear choice for batch work. At $0.05 per image, you can afford to generate aggressively and curate after. Reserve Pro and GPT-Image for targeted generation where material accuracy or photorealism justifies the per-image premium.
For reference, Burnish Pro's plans:
- Free tier: 150 one-time credits (30 Flash images to test the workflow)
- Pro: $39/month, 2,000 credits (400 Flash images, or 4 full batches of 100)
- Enterprise: $99/month, 8,000 credits (1,600 Flash images)
A Pro subscriber can run a 100-image batch every few days and still have credits left for variants, refinement, and 3D modeling.
How Batch Processing Works
Burnish Pro processes batches in groups of roughly 20 images at a time. When you submit a run of 100 designs, the system queues them into sequential batches and processes them through the AI provider.
The practical timeline:
- First images appear within about 60 seconds
- 20-image batch completes in roughly 60 seconds
- Full 100-image run takes approximately 5 minutes
You see results as each batch finishes — there is no waiting for the entire run to complete before reviewing output. Start curating the first 20 results while batches 2 through 5 are still processing.
Runs support anywhere from 1 to 100 designs per submission. You set the count when configuring the run, along with the jewelry type, style parameters, metal, stones, and photography preset.
When Batch Generation Makes Sense
Batch runs are a volume tool. They are built for workflows where you need to explore a wide design space quickly and filter down to the strongest results.
Seasonal collections. Generate 100 variations of "holiday tennis bracelet" across different metals, stone combinations, and style treatments. In 5 minutes you have a raw collection to curate from — instead of spending days sketching options or hours building CAD concepts.
Client exploration. A client says "I want an Art Deco emerald ring." That brief could mean a hundred different things. Generate 50 variations interpreting the brief differently — geometric halos, step-cut clusters, channel-set baguettes, filigree gallery work — and let the client point to what resonates. The cost of 50 Flash images is $2.50.
Line sheet generation. Manufacturers building a new product line need dozens of concepts to evaluate before committing to CAD and prototyping. A batch run produces a complete line sheet draft in minutes.
Style exploration. Trying to define a new aesthetic direction? Run 100 images with a consistent base (say, 14K yellow gold pendant) and vary the style parameter across every option — minimalist, Art Deco, bohemian, Gothic, mid-century modern, organic. Map the style space visually instead of imagining it abstractly.
Trade show prep. Generate 100 concepts for a booth display or catalog, curate the top 20, then run 4-angle variants on the 10 hero pieces. Total cost for the hero set with variants: 100 × 5 (initial) + 10 × 4 × 5 (variants) = 700 credits = $7.00.
Configuration Strategy for Batch Runs
Batch runs amplify whatever you configure. Minimal selections generate 100 wildly different results. Over-specified selections generate 100 near-identical images. The sweet spot is locking the fundamentals and leaving room for the AI to vary the design details.
Lock the fundamentals. Select the jewelry type, metal, and primary stone. These define the category and stay consistent across all 100 outputs.
Leave room for variation. Do not over-specify style or stone patterns unless you want tight consistency. Selecting a single style like Art Deco keeps the collection cohesive. Selecting multiple styles or leaving style open gives the AI more creative range.
Set the photography preset. Use a single preset across the batch so all images look like they belong in the same catalog. Burnish Pro's 6 presets — Clean E-Commerce, Lifestyle, Editorial Dark, Natural Light, Studio Dramatic, and Auto — control lighting, background, and composition consistently across the full run.
Use Custom Instructions for specifics. If you need a particular detail across all outputs — "oval center stone, diamond accents" — put it in the Custom Instructions field rather than trying to force it through selector combinations alone.
A well-configured batch might be: Ring > 18K White Gold > Natural Sapphire (oval cut) > Diamonds as accent, with photography set to Clean E-Commerce and quantity at 100. Specific enough to stay on-brief, open enough to produce genuine variety.
For designers who want full control, each main category (Type, Style, Materials) offers a Custom Prompt override — but the form-based workflow produces expert-quality results without it.
For the full workflow from concept to 3D, see the complete AI jewelry design guide.
What to Do With 100 Results
The curation step is where design expertise matters. Generating 100 images is fast; the value is in what you do with them.
First pass: eliminate. Scroll through all 100 and discard anything with obvious issues — broken geometry, unrealistic proportions, stones that look pasted on. This typically cuts the set by 30-50%.
Second pass: categorize. Group the survivors by style direction. You will often find that the AI clustered around 3-5 distinct interpretations of your configuration. These clusters are your design directions.
Third pass: select hero pieces. Pick the 5-10 strongest images across clusters. These are candidates for variant generation (4-angle views) or 3D mesh creation.
This three-pass process takes 10-15 minutes for a 100-image batch. Combined with the 5-minute generation time, you go from a blank form to a curated shortlist in under 20 minutes.
When NOT to Batch
Batch generation is the wrong tool when you need precision control over a single design.
If a client has approved a concept sketch and you need an exact render matching specific dimensions and proportions, use individual generation with a detailed configuration and iterate one image at a time. Adjust your selections, regenerate, refine. The per-image workflow gives you control that batch runs intentionally sacrifice for volume.
Similarly, if you are refining an existing design — adjusting the band width, changing a stone shape, modifying a setting detail — use the studio refinement tools rather than generating a fresh batch. Refinement preserves the original design intent. Batching explores new territory.
The two workflows complement each other:
| Workflow | Speed | Control | Best For |
|---|---|---|---|
| Batch (1-100 images) | 100 images in 5 min | Low — AI interprets freely | Exploration, collection concepting |
| Individual + Refine | 1 image at a time | High — iterate on specifics | Approved concepts, client revisions |
| Variants | 4 angles per design | Medium — consistent views | Presentations, manufacturing review |
From Batch to Production
The typical path from a batch run to a manufacturable design:
- Batch generate 50-100 concepts (5 min, $2.50-$5.00)
- Curate to 5-10 hero pieces (10 min)
- Generate variants for hero pieces — 4 angles each (2 min, $1.00-$2.00)
- Create 3D base meshes for top picks — 60 credits each (varies, $0.60/model)
- Hand off base meshes to a CAD artist for production refinement
The AI handles steps 1-4. Human expertise takes over at step 5 for the precision work that manufacturing demands. For the full pipeline breakdown, see From 2D Render to 3D Print.
This is not a replacement for the design and manufacturing skills that make jewelry production possible. It is a way to front-load the ideation phase so those skills are applied to the strongest concepts rather than the first idea that came to mind.
For a beginner-friendly introduction to the platform, start with How to Design Jewelry with AI.
Cost Planning for Teams
For teams running batch generation regularly, here is how credit consumption maps to real usage:
| Activity | Weekly Volume | Credits/Week | Monthly Credits |
|---|---|---|---|
| 2 batch runs (50 images each) | 100 images | 500 | 2,000 |
| Variants on 10 hero pieces | 40 variants | 200 | 800 |
| 5 individual refinements | 5 images | 25-100 | 100-400 |
| 2 base 3D models | 2 models | 120 | 480 |
| Total | 3,380-3,680 |
That workload fits within an Enterprise plan (8,000 credits/month at $99). A lighter workload — one batch run per week with occasional variants — would fit within the Pro tier (2,000 credits/month at $39). Need a burst of extra capacity for a seasonal push or a large client order? One-time credit top-ups are available on paid plans — buy what you need without changing your subscription. Full plan details are on the pricing page.
Generate your first batch of jewelry concepts. Start free with 150 credits — enough for 30 Gemini Flash images, no credit card required.