Short answer
Use a single “anchor” prompt + seed + camera spec (lens, height, tilt, view) and fix aspect ratio. For lighting iterations, keep the anchor constant and only change the lighting phrase or swap a lighting reference image. If you need pixel‑perfect identical camera angles across many renders, use a 3D render or a photographed plate—Midjourney can be very consistent visually but will not guarantee exact geometry every time.

Recommendation
Create one anchor image with an explicit camera block in your prompt, save its seed and image, then reuse that exact seed + (preferably) the uploaded anchor image for all variations. Use concise, repeatable camera wording (e.g., “50mm, 3/4 view, camera height 1.1m, 20° downward tilt, eye-level product framing”). Iterate lighting by changing only the lighting clause or swapping in different lighting reference photos.

Decision criteria (which approach to pick)
- Use Midjourney when: budget or time is limited, you want quick photoreal explorations, and you can accept small geometric differences between images.
- Use 3D software or a product photoshoot when: you require pixel‑exact camera angles across hundreds of renders, need production-ready 3D assets, or have a brand/engineering requirement for exact dimensions.
- Use Midjourney + 3D combo when: you can export a simple clay render or turntable then use MJ to stylize lighting/finish while keeping angle exact.

Practical checklist (step-by-step)
1) Anchor shot: craft the anchor prompt and generate a final you like. Include camera block and framing explicitly. Example template:
product shot of [PRODUCT], 3/4 view, camera: 50mm lens, camera height 1.1m, 20° downward tilt, tight crop, studio seamless background, photorealistic --ar 4:5 --seed 123456 --v 5 --q 2 --stylize 50
2) Save the seed and the generated image ID. Also save the exact prompt text.
3) Reuse anchor image: upload the anchor image and reference it in the prompt (image URL first). Use the same seed and same framing text to lock camera feel.
4) Lighting iteration method A (text): keep everything identical but change only the lighting clause: “softbox key left, 45°, warm 3200K, silver rim light right” vs “hard top spotlight, high contrast, cool 5600K.” Generate variations.
5) Lighting iteration method B (reference): upload 1–3 lighting reference photos and include them before your anchor in the prompt to nudge lighting style while keeping camera.
6) Evaluate and refine: use MJ’s variation buttons on the anchor output if you want small changes, or re-run with the same seed for consistency. If outputs drift, re-upload anchor and repeat.
7) Batch: automate prompt templates (CSV) and run in small batches. Use ChatGPT to produce consistent prompt permutations if helpful.

Best‑for / Avoid‑if
- Best for: fast, varied photoreal explorations of product finishes and lighting directions; small catalog shoots when absolute geometry tolerance is relaxed.
- Avoid if: you need exact pixel alignment across images, mechanical orthographic views, or exact CAD‑matching—use 3D or photography instead.

Notes on budget & team
Midjourney is low cost and fast for one–two designers or small teams. For large catalogs or production pipelines (many SKUs, exact angles), invest in a 3D pipeline or a single controlled photoshoot. Use ChatGPT to generate prompt variations and keep templating errors out of your batch runs.

If you'd like, I can draft a reusable prompt template and three lighting variants tailored to your product (material and scale).

Compare Midjourney and Leonardo AI