The email arrived at 6:42 PM on a Thursday. Subject line: "Shoot files — prop missing." The art director had just received the final selects from a product photography session for a premium candle brand. The photos were beautiful — soft directional light, cream linen backdrop, the candle positioned at a perfect angle. But the brief had called for a specific prop: a vintage brass candle snuffer, placed just to the left of the candle, with a subtle reflection on the surface.
The snuffer had been on-set. But in the selects, it was absent — the photographer had removed it for a clean-product shot and forgotten to shoot the variant with the prop. The retoucher who could comp one in was booked for two weeks. A reshoot would cost $2,800 for the half-day studio minimum, plus re-coordination with the prop house that had rented out the snuffer.
Total damage: a missing brass object was about to cost nearly three thousand dollars and delay a campaign launch by two weeks. Or we could try fixing it with Lovart.
The Problem, Precisely Defined
The delivered photo showed:
- A matte black candle in a ceramic vessel, centered on a cream linen surface.
- Soft window light from the left.
- Clean composition with generous negative space on the left side — exactly where the brass snuffer was supposed to sit.
- No snuffer. No reflection. No metal interaction with the light.
The missing prop wasn't a decorative flourish. It was part of the brand story — the candle brand positioned itself as "ritual luxury," and the brass snuffer was the ritual object. Without it, the photo was just a candle. With it, the photo was an invitation to a practice.
The Solution: Generate, Composite, Match Lighting
Step 1: Generate the missing prop.
We described the snuffer as precisely as possible into Lovart's ChatCanvas:
"A vintage brass candle snuffer, bell-shaped with a slender curved handle. Polished brass with slight patina and tarnish — authentic aged look, not factory-new. The snuffer is resting on a cream linen surface. Viewed from a 45-degree angle. Soft window light coming from the left, casting a gentle shadow to the right. The snuffer should be approximately 4 inches tall in the composition. Isolated object with no background — alpha transparency."
Lovart generated the snuffer as an isolated transparent PNG. The brass material was convincing — warm gold tones with subtle oxidation spotting. The proportions matched the reference measurements we'd added to the prompt. The lighting direction (left-side soft window) matched the original photo's lighting.
Step 2: Composite into the original photo.
We opened the original candle photo in Lovart and used the canvas as the base layer. Imported the generated snuffer as an overlay. Positioned it on the left side of the composition, approximately 3 inches to the left of the candle, matching the art director's original layout sketch.
The snuffer sat naturally in the composition — the empty negative space on the left had been designed for it. The composition went from "clean product shot" to "ritual still life" with one element addition.
Step 3: Match lighting and shadows.
This is the step where most composites fail. The generated prop had left-side lighting, matching the original photo's direction. But the intensity was slightly off — the snuffer was brighter than the candle, creating a subtle but noticeable mismatch in light levels.
Touch Edit: selected the snuffer → "Reduce brightness by 15%. Increase contrast slightly to match the lighting environment of the candle in the same frame. Add a soft drop shadow cast to the right — same softness and opacity as the candle's existing shadow."
The shadow adjustment was the critical detail. The candle cast a soft shadow from the left-side window light. The snuffer needed an identical shadow behavior — same direction, same softness, same opacity — or it would look like a paste-in. Lovart's shadow generation analyzed the existing shadow in the scene and produced a matching cast shadow for the inserted element.
Step 4: Color grade for cohesion.
Final pass: a unified color grade that treated the original photo and the generated prop as a single image. We applied a subtle warm tone curve (lifted blacks, slightly compressed highlights) across the entire composite. This is the "glue" step — the color grade makes disparate elements read as a single photograph captured under the same conditions.
Result: a seamless product photo with the missing prop. Total time: 12 minutes. Total cost: zero additional dollars. The art director couldn't tell which element had been generated and which had been photographed.
The Economics
| Cost Item | Reshoot Approach | Lovart Approach | |------------|-----------------|-----------------| | Studio half-day minimum | $2,800 | $0 | | Prop house re-rental | $200 | $0 | | Photographer fee | Included in studio | $0 | | Retoucher fee | $400 | $0 | | Campaign delay cost | ~$5,000 (2 weeks lost) | $0 | | Total | $8,400 | $0 (included in subscription) |
The $5,000 delay cost is estimated — two weeks of delayed ad spend at the brand's daily budget, with creative that would have performed worse without the ritual-prop storytelling element. The actual cost of delay depends on your campaign economics, but the principle holds: a missing prop shouldn't be a campaign-level problem.
When This Approach Works
- The missing element is a discrete object (prop, accessory, decorative element) rather than a fundamental part of the main subject.
- The original photo has clear negative space where the element belongs — you're filling a gap, not overwriting existing content.
- Lighting direction is consistent or can be described consistently in the generation prompt.
- The element is something AI renders well: metal objects, pottery, textiles, plants, books, candles, bottles.
When It Doesn't Work
- The missing element is a person, a face, or a hand — AI generates people inconsistently, and compositing a generated person into a real photo is extremely difficult to do convincingly.
- The element needs to integrate with complex reflections (mirrors, water surfaces, curved glass).
- The element casts complex shadows across the main subject (the shadow overlay becomes technically challenging).
| Image | Description | Placement | |-------|-------------|-----------| | original-photo-no-snuffer.jpg | The delivered product photo — candle only, empty left side | Introduction | | generated-snuffer.png | The AI-generated brass snuffer on transparency | Step 1 | | composite-no-shadows.jpg | Snuffer placed in composition, before shadow matching | Step 2 | | final-composite-with-shadows.jpg | Complete composite with matched shadows and color grade | Step 3–4 | | side-by-side-comparison.jpg | Original photo vs. final composite — spot the difference | Conclusion | | cost-comparison-table.jpg | Visualized cost breakdown: reshoot vs. Lovart fix | Economics |
FAQ
Can I fix a photo where the main product itself has a flaw, not just a missing prop? Yes, with Touch Edit. Select the flawed area (a scratch, a dent, a label misalignment) and describe the correction. For major flaws on the main subject, generate the correction at high resolution and inspect the result carefully — flaws on the focal point are more noticeable than flaws on peripheral elements.
What if the lighting in my photo is complex — multiple sources, colored gels, unusual angles? Describe the lighting in the prompt with as much detail as you can. "Soft window light from upper left, warm color temperature, with a subtle fill from a white bounce card on the right." The more specific the lighting description, the better the generated element will match. If matching is imperfect, use Touch Edit for post-composite lighting adjustment.
Does the generated element hold up at print resolution? For high-quality print (300 DPI, magazine or catalog), generate at Lovart's maximum resolution (4K) and inspect at 100% zoom. AI-generated objects can show subtle texture inconsistencies that are invisible on screen but noticeable in print. For e-commerce and digital advertising, any resolution above 2K is typically sufficient.
Can I use this technique to generate product variants — different colors, materials, sizes? Yes. This is actually one of the highest-value use cases. Shoot one product (the hero color), then generate variants in other colors or materials using the same process — describe the variant, generate it as an isolated object, composite it into the original photo. One shoot produces visual content for an entire SKU line.
What about reflections? How do I match reflective surfaces? Reflections are hard. If your original photo has a reflective surface (mirror, polished floor, glossy table), the generated element needs a matching reflection. Prompt for: "Include a subtle reflection of the object on the [describe surface] below, matching the reflectivity of the surface." For imperfect results, add the reflection manually in post-processing rather than expecting the AI to nail it in one generation.
Is this ethically different from photoshopping a product to look better than it is? Yes. Adding a missing prop is compositional correction — restoring the intended scene. Altering the product itself to misrepresent its features or quality is deceptive. The ethical line is: are you fixing a production oversight, or are you creating a false representation of what the customer will receive? Prop addition is the former. Product enhancement is the latter.
Can I use Lovart to fix multiple issues in a single photo simultaneously? Yes. Process each fix independently — generate the missing prop, composite it, then move on to the next issue. Don't try to fix everything in one generation, because the model will struggle with multi-problem prompts. Sequential small fixes produce better results than one large fix attempt.
Internal Links
- Ugly to Lovely — We Took a Bad MS Paint Drawing and Let AI Fix It
- Isolating Objects — How to Turn AI-Generated Items into Transparent Stickers
- The Real Estate Rescue — Removing a Garbage Can from a Generated House Listing
- Prompting for Repairs — What Words Actually Work When Asking AI to Fix a Mistake
Laura Keating is a commercial photographer and creative producer who has managed product shoots for DTC brands in home goods, beauty, and fashion. She has produced over 150 product photography sessions and has experienced every type of shoot-day failure — from missing props to broken products to weather-ruined location shoots. She began using Lovart in 2025 for shoot rescue and post-production fixes, and estimates it has saved her clients over $40,000 in reshoot costs.
