A patient wants teeth that are "whiter and straighter, but not fake," and you have a shade guide, a mirror and your own words to explain what six veneers would look like. An AI preview made from one photo can show a version of that smile before the consult ends. It can also show a smile the patient's teeth, gums and bite cannot support.
Below: what studies say about previews and case acceptance, how close simulations come to deliverable work, what a 2D image cannot show, which consent and privacy steps to take, and how to turn an approved preview into lab instructions.
Key Takeaways
- In a randomized trial of 60 patients, those shown a smile simulation scored higher on understanding, confidence and willingness to start orthodontics first, measured right after the visit.
- Of 252 survey participants, 63.2% mistook smile images from a general-purpose AI generator for real results, and AI images averaged 69.2 against 53.9 out of 100 for attractiveness.
- In a 33-patient study, patients chose the AI design 69.7% of the time and prosthodontists 51.5%, with no agreement beyond chance.
- One AI simulator widened the lateral incisors in 62.7% of 51 simulations, more than tooth movement alone could deliver.
- In vitro, trial restorations deviated from a 3D smile design by 0.11 to 0.21 mm on average.
- In a mathematical model, 12-inch selfies make the nose look about 30% larger: close-range photos distort proportions.
Quick Answer
Smile simulation helps patients picture a change; in short-term studies, clinician-made simulations raised understanding, confidence and stated willingness; one-photo AI previews have weaker evidence. It is a communication aid, not a treatment plan. A 2D image can alter tooth size, shade and gum display but cannot show occlusion, periodontal limits or how much tooth must be prepared. Show it after the exam, label it a simulation, record consent, and confirm the design in the mouth before irreversible work.
Do Smile Previews Improve Case Acceptance?
Understanding and confidence go up
In a 2026 randomized trial, 60 patients seeking a smile makeover had a conventional consultation or one with a scan-based simulation video. The simulation group scored higher on understanding, confidence and willingness to have orthodontic treatment before restorations. The authors call these immediate impressions and urge caution.
In a 2026 single-group study of 80 esthetic patients, confidence in the final smile rose from 3.0 to 4.6 on a 5-point scale after a non-AI simulation, and 85% accepted treatment. There was no control group, and the reasons of the 15% who declined were not analyzed.
A picture alone does not close the case
In Madrid, 51 people saw an AI simulation of their own smile. Their smile rating rose from 5.84 to 7.00 out of 10, yet intent to start orthodontic treatment fell from 49.0% to 45.0%. The authors saw limited influence on motivation without clinical guidance.
Reviews are positive but thin: a 2025 meta-analysis of AI-based smile design pooled 6 small studies, including single-patient cases, into 58% satisfaction, with high heterogeneity and publication bias.
| Study | Participants | Design | Main finding | Caveat |
|---|---|---|---|---|
| Ali et al., 2026 | 60 patients | Randomized: simulation vs. conventional consultation | Higher understanding, confidence and willingness | Immediate, self-reported |
| Qadiri et al., 2026 | 80 patients | Before and after a simulation | Confidence 3.0 to 4.6 of 5; 85% accepted | No control group |
| Mourgues et al., 2026 | 51 people | AI simulation of own smile | Intent to treat 49.0% to 45.0% | One app, one city |
| Almohareb et al., 2026 | 33 patients, 20 prosthodontists | AI vs. expert design, rated blind | Patients chose AI 69.7%, prosthodontists 51.5% | Virtual designs only |
| Alqahtani et al., 2026 | 180 raters | AI vs. dentist design | Dentist designs judged more natural (76.1% and 78.3%) | Two patients only |
How Close Is a Simulation to What You Can Deliver?
AI designs and expert designs
Results conflict. In the 33-patient study, automated designs scored better on a dental esthetic index, but their central incisors were slightly wider and longer (0.29 mm and 0.32 mm) and their buccal corridors narrower. Patients and prosthodontists picked the same design only 54.5% of the time. In the 180-rater study, dentist designs scored higher, and raters spotted the AI design in 66.1% and 69.4% of cases.
From design to try-in
In a lab study, trial restorations made by four digital workflows deviated from the 3D smile design by 0.11 mm (milled) to 0.21 mm (printed index) on average. In a 20-patient trial, 80% found the digital workflow more convenient, yet 65% chose the conventional wax-up after trying both mock-ups.
2D photos and 3D scans disagree
In 30 patients, lateral incisor and canine widths measured differently in photos than in scans, and the 2D plan differed from the 3D plan in every linear measurement except incisor height, although the final veneers matched both. A 2025 scoping review found 3D designs more accurate than 2D.
What Can a 2D Smile Image Not Show?
An AI preview edits pixels. It does not know bone level, bite or enamel thickness.
| Preview shows | What it cannot tell you | What to check or send |
|---|---|---|
| Wider teeth | Where the width comes from: space, reduction or tooth movement | Measure spaces on the scan; plan widths tooth by tooth |
| Longer teeth | Whether length is incisal (bite, guidance, speech) or gingival | Incisal edge position in mm; periodontal evaluation |
| Less gum display | Bone level, need for crown lengthening | Probing, radiographs, perio consult |
| Whiter teeth | A real shade under clinical light | Shade tab photo, written shade, stump shade for e.max |
| Narrow buccal corridors | Whether teeth can move or be built out | Arch scan; orthodontic opinion |
| Centered midline | Whether the dental midline can move | Full-face photo; keep or correct |
Proportions drift toward a template
In a study of a commercial AI orthodontic simulator, 62.7% of 51 simulations widened the lateral incisors (21.6% the centrals), which would need restorative work as well as tooth movement. It centered an off-center lower midline less often than an upper one (65% against 77.9%). The authors warn it can change proportions without considering the patient's anatomy.
Gums, bone and biologic limits
A preview lengthens teeth at the gingival margin as easily as at the incisal edge. Clinically that may mean crown lengthening, and the 2017 World Workshop consensus associates restorative margins within the supracrestal connective tissue attachment with inflammation or loss of periodontal support.
Bite, reduction and shade
A frontal photo shows no occlusion, guidance or parafunction, and cannot tell whether a change is additive or needs reduction. The preview's white is set by software and the patient's screen. Even measured photos disagree: in a 2026 lab comparison, DSLR and smartphone photos and two spectrophotometers all differed from a reference instrument beyond the acceptability threshold. See our guide to shade matching with photos.
How Do You Keep Expectations Realistic?
Share of 252 survey participants who mistook AI-enhanced smile images, made with a general-purpose image generator, for real post-treatment photos (Gasparello et al., 2026).
Participants also rated the AI images more attractive than real results, and the authors warn that AI smile simulations may create unrealistic expectations. Laypeople judge previews generously: in a Zurich study of AI video mock-ups, lay participants rated their mock-ups higher than dentists did on every item except tooth alignment and facial profile.
- Show the preview after the exam and radiographs.
- Call it a simulation of a possible result, not a guarantee.
- Name the treatment behind it, including any orthodontic or periodontal step it assumes.
- Point out anything you cannot deliver, then edit or redo the preview.
- In advertising, the ADA Code asks dentists to avoid statements likely to create an unjustified expectation about results.
What Should Consent and Documentation Cover?
The ADA Code asks dentists to inform patients of the proposed treatment and reasonable alternatives so they can take part in decisions. A preview supports that conversation; it does not replace it.
Photo consent
Get written consent that names each use: the record, the lab, sending the image to the patient, and marketing. HIPAA requires written authorization to use or disclose protected health information for marketing, with narrow exceptions.
A neutral privacy checklist
- Where are the photo and preview stored, for how long, and who can delete them?
- Who can open a shared link, and does it expire?
- Will a vendor create, receive, maintain or transmit patient health information for you? Under HIPAA, such a vendor is generally your business associate, requiring a written agreement. Ask whether the vendor signs one before you upload patient photos, and involve your compliance advisor.
- Do staff use personal phones? Decide where those files go.
Documentation
Save the exact image the patient saw, with the date and tool, plus what the patient approved (length, shape, shade, which teeth) and what you said it could not show. Record requests you declined, and save each new version instead of overwriting.
How Do You Turn an Approved Preview Into Lab Instructions?
An approved preview shows the direction the patient liked. The lab still needs measurements and references:
- Photos: full face at rest and smiling, a retracted frontal view and a shade tab photo, taken from a distance, not at selfie range.
- Shade: the target shade and guide system in writing, and the stump shade for e.max. "Like the preview" is not a shade.
- Midline: where the facial midline falls, and whether to keep or correct the dental midline and any cant.
- Incisal edge: central incisor display at rest in millimeters, and how much length to add or remove.
- Smile line: how the incisal edges follow the lower lip, and gingival display on a full smile.
- Proportions: which teeth are included, with width or length changes tooth by tooth.
- Limits: anything in the preview you did not prescribe, such as wider laterals, so it is not copied.
Choosing a material? Compare e.max, zirconia and feldspathic veneers.
Elegant Smile: Our Lab's Preview App
Elegant Dental Laboratory in Brooklyn built Elegant Smile for US dentists and clinics. It turns one patient photo into an AI smile preview in about 30 seconds, and the preview is sent to the patient by a private link. It starts with 10 days unlimited, with no card required. The limits above apply to it as to any preview made from a photo. Its terms state that Elegant Smile is not a HIPAA-covered service and does not sign business associate agreements, so the practice obtains any patient consent needed before uploading a photo.
Esthetic Cases at Elegant Dental Laboratory
- Every case is made in house in Brooklyn by more than 35 technicians, and digital scans are reviewed before production.
- IPS e.max Press veneers from 0.3 mm and layered crowns; zirconia in solid 3Y, 9-layer 5Y and layered options, following the brand or grade on the Rx.
- STL and PLY files from all major scanners (submit a digital case); 5 business day standard turnaround on most restorative cases, rush options, and up to a 7-year warranty on final restorations.
- Free pickup and delivery by our own drivers in all five New York City boroughs and 13 northern New Jersey counties, and UPS labels anywhere (send a case). Call 877-335-5221.
FAQ
Do AI smile previews increase case acceptance?
Clinician-made simulations raised short-term understanding, confidence and stated willingness; one AI app study did not.
Can I send the AI preview to the lab as the design?
Send it as a reference. The lab needs scans, photos, a written shade and measurements such as incisal edge position and midline.
What can't a smile simulation show?
Occlusion, guidance, periodontal and bone limits, the reduction a change needs, and the true shade.
Do I need consent for smile photos?
Written consent naming each use is good practice; HIPAA requires written authorization for marketing, with narrow exceptions. Confirm with counsel.
Are AI designs as good as an expert's?
Studies disagree, so review every design before you show it.
Sources (21)
- Ali A, Abdulateef D, Barzinji L. Effect of digital smile simulation on patients' decision to accept combined orthodontic restorative treatment planning for smile makeover. Sci Rep. 2026 (published online July 15, 2026). doi:10.1038/s41598-026-62296-7
- Qadiri SY, et al. Impact of Digital Smile Design on Patient Satisfaction and Treatment Acceptance in Esthetic Dental Practice: A Clinical Study. Cureus. 2026;18(4):e106350. doi:10.7759/cureus.106350
- Mourgues T, et al. A Prospective Longitudinal Observational Study on the Emotional Impact of AI-Simulated Smiles on Orthodontic Patient Motivation. Int J Dent. 2026;2026:8425551. doi:10.1155/ijod/8425551
- Saini RS, et al. Impact of artificial intelligence-based digital smile design on patient and clinician satisfaction and facial esthetic outcomes: A systematic review and meta-analysis. Digit Health. 2025;11:20552076251388392. doi:10.1177/20552076251388392
- Almohareb T, et al. Clinical and Patient Comparison of AI and Expert Digital Smile Design: A Prospective Paired Study. Dent J (Basel). 2026;14(3):166. doi:10.3390/dj14030166
- Alqahtani SM, et al. Comparative perception of smiles designed by AI vs. dentist using standard esthetic rules: a visual preference study among laypeople and dental professionals. Odontology. 2026 (published online August 21, 2026). doi:10.1007/s10266-026-01533-x
- Taha D, Allam S, Morsi T. Accuracy of computer-aided design trial restorations fabricated with different digital workflows. J Prosthet Dent. 2024;132(3):578-585. doi:10.1016/j.prosdent.2023.09.034
- Mocelin RC, et al. Assessment of patient and dentist preference between conventional and digital diagnostic waxing. Int J Esthet Dent. 2021;16(3):300-309. PubMed 34319665
- Ortensi L, et al. Digital planning of composite customized veneers using Digital Smile Design: Evaluation of its accuracy and manufacturing. Clin Exp Dent Res. 2022;8(2):537-543. doi:10.1002/cre2.570
- Baaj RE, Alangari TA. Artificial intelligence applications in smile design dentistry: A scoping review. J Prosthodont. 2025;34(4):341-349. doi:10.1111/jopr.14000
- Mourgues T, et al. Artificial Intelligence in Aesthetic Dentistry: Is Treatment with Aligners Clinically Realistic? J Clin Med. 2024;13(20):6074. doi:10.3390/jcm13206074
- Jepsen S, et al. Periodontal manifestations of systemic diseases and developmental and acquired conditions: Consensus report of workgroup 3 of the 2017 World Workshop on the Classification of Periodontal and Peri-Implant Diseases and Conditions. J Periodontol. 2018;89 Suppl 1:S237-S248. doi:10.1002/JPER.17-0733
- Türkoğlu Tarı E, Bayraktar Y. Agreement among tooth color measurement devices: An experimental comparison. Am J Dent. 2026;39(3):134-138. PubMed 42247363
- Ward B, et al. Nasal Distortion in Short-Distance Photographs: The Selfie Effect. JAMA Facial Plast Surg. 2018;20(4):333-335. doi:10.1001/jamafacial.2018.0009
- Gasparello GG, et al. Human perception of AI-generated post-treatment orthodontic facial images: factors associated with misclassification. Clin Oral Investig. 2026;30(8). doi:10.1007/s00784-026-07020-5
- Pachiou A, et al. Perception of Orofacial Esthetics Using a Scan-Free Virtual Video Mock-Up: A Comparative Analysis of Self-Evaluation Versus Unknown-People Assessment. J Esthet Restor Dent. 2026;38(3):652-662. doi:10.1111/jerd.70059
- American Dental Association. Principles of Ethics and Code of Professional Conduct, with official advisory opinions revised to October 2025 (Sections 1.A and 5.F.2). PDF
- 45 CFR 160.103, definition of business associate. eCFR
- 45 CFR 164.502(e), disclosures to business associates and the written agreement. eCFR
- 45 CFR 164.508(a)(3), authorization required for marketing. eCFR
- 45 CFR 164.514(b)(2), identifiers removed for de-identification. eCFR
Reviewed by Gary Fingerman, founder and CEO of Elegant Dental Laboratory, Brooklyn, New York (founded 2007). Last reviewed: October 3, 2026. Study values are from published research; this article is not legal advice, so confirm privacy and consent requirements with your compliance advisor.
