Why outpainting
wins.
A controlled test of three ways to reformat vertical UGC for the TV screen, and why the method you choose is a creative decision, not a packaging afterthought. 300 US women, one ad, three framings. Outpainting won six of seven metrics and was the only format that looked native to the screen.
The same Caraway UGC ad, reframed three ways to fill a 16:9 TV screen.
The short
version.
Most UGC is shot vertical or square. Putting it on Connected TV means filling a 16:9 frame, and brands typically reach for one of three fixes: letterboxing (black bars around the original), side-blur (the gaps filled with a blurred blow-up of the video), or outpainting (AI extends the scene so it fills the whole canvas). We tested whether that choice actually matters.
Using a single Caraway cookware UGC ad, we produced all three versions with identical content, changing only the framing treatment, and showed each to a separate, randomized group of US women 25-45 (n = 300) on Pollfish. Across nine measures of enjoyment, attention, purchase intent, brand perception, production value, and platform fit, outpainting won six of seven scored metrics and the overall composite (3.46 vs. 3.23 vs. 3.16 on a 1-5 scale). More importantly, it was the only format to win, by statistically significant margins, on the two questions that define the whole problem: "looks like it was made for TV" and "looks professional."
The takeaway is simple. How you fill the frame is not cosmetic. Letterboxing and side-blur leave UGC looking like phone footage stretched onto a television. Outpainting makes the same footage read as native, professional CTV inventory, and that perception carries through to attention, brand favorability, and purchase intent.
scored metrics won by outpainting, including the overall composite (3.46 vs. 3.23 vs. 3.16).
said the outpainted cut looks "made for TV," versus about 38% for the others. Statistically significant (p = 0.002).
called it "professional" or better, versus 36% letterboxed. Also statistically significant.
How we
tested.
A between-subjects design: each respondent was randomly assigned to one of the three formats and answered only for the version they saw, so no one could compare treatments side by side. This isolates the framing method as the single variable. The underlying creative, audio, brand, and message were identical across all three cells.
| Format | Respondents | What it is |
|---|---|---|
| Letterboxed | 115 | Original vertical/square video centered with black bars top and bottom. |
| Side Blur | 101 | Original centered, gaps filled with a blurred enlargement of the video. |
| Outpainted | 84 | Frame AI-extended so the scene fills the full 16:9 canvas. |
Sample. n = 300 US women aged 25-45 (Caraway's core buyer), fielded via Pollfish on 2026-06-23.
Measures. Nine survey items spanning gut reaction, skip-resistance, attention, purchase intent, brand perception, production value, and platform fit.
Scoring. Every 5-point scale is mapped to 1-5, where 1 = the lowest anchor ("Not at all" / "Very unlikely" / "Poor" / "Looks amateur") and 5 = the highest ("A lot" / "Very likely" / "Excellent" / "Looks very professional"). Higher is better everywhere. The one behavioral item (skip behavior) is scored 1-3.
Significance. Pairwise two-sided Mann-Whitney U tests. At about 84 to 115 per cell, only sizable gaps clear p < 0.05; those are flagged Sig. Other differences are reported as directional: real-looking leans the sample isn't large enough to certify on their own.
Reading directional results. No single brand-intent metric is significant in isolation. But all five point the same way, toward outpainting. Five independent coin-flips landing on the same side is itself the signal.
Question
by question.
Overall, how much did you enjoy this video?
Not at all (1) → A little (2) → Somewhat (3) → Quite a bit (4) → A lot (5)
| Format | Not at all | A little | Somewhat | Quite a bit | A lot | Avg |
|---|---|---|---|---|---|---|
| Letterboxed | 12% | 21% | 18% | 24% | 24% | 3.28 |
| Side Blur | 11% | 14% | 27% | 16% | 33% | 3.46 |
| Outpainted | 7% | 18% | 26% | 23% | 26% | 3.43 |
Takeaway: On raw enjoyment, anything that fills the frame beats black bars. Outpainted and Side Blur tie at the top while Letterboxed trails. But gut reaction is a near-wash. The real separation between formats shows up the moment you ask anything that matters.
If this appeared before a video you wanted to watch, what would you do?
Behavioral. Watch-score = Skip (1) / Watch some (2) / Watch the whole thing (3).
| Format | Skip ASAP | Watch some | Watch the whole thing | Not sure | Watch-score |
|---|---|---|---|---|---|
| Letterboxed | 22% | 26% | 49% | 3% | 2.28 |
| Side Blur | 25% | 24% | 50% | 2% | 2.25 |
| Outpainted | 17% | 32% | 51% | 0% | 2.35 |
Takeaway: Outpainted is the hardest to skip. It posts the lowest "bail immediately" rate (17% vs. 22-25%) and the highest watch-through. In a pre-roll context, that's fewer wasted impressions before the message even lands.
How much did this video hold your attention?
Not at all (1) → A little (2) → Somewhat (3) → Quite a bit (4) → Completely (5)
| Format | Not at all | A little | Somewhat | Quite a bit | Completely | Avg |
|---|---|---|---|---|---|---|
| Letterboxed | 13% | 30% | 13% | 23% | 21% | 3.10 |
| Side Blur | 15% | 20% | 17% | 24% | 25% | 3.24 |
| Outpainted | 8% | 18% | 26% | 26% | 21% | 3.35 |
Takeaway: Outpainted holds attention best and loses the fewest viewers outright. Just 8% said it didn't hold them at all, versus 13-15% for the others. Letterboxing sheds the most to the "barely watched" tail.
How interested are you in learning more about Caraway?
Not at all (1) → A little (2) → Somewhat (3) → Quite a bit (4) → Very (5)
| Format | Not at all | A little | Somewhat | Quite a bit | Very | Avg |
|---|---|---|---|---|---|---|
| Letterboxed | 21% | 18% | 23% | 17% | 20% | 2.97 |
| Side Blur | 18% | 23% | 19% | 12% | 29% | 3.11 |
| Outpainted | 12% | 20% | 23% | 23% | 23% | 3.24 |
Takeaway: Outpainted nearly halves the "not interested at all" rate, 12% versus 21% for Letterboxed. Fewer slammed doors means more of the audience is actually reachable.
Next time you're shopping for cookware, how likely are you to consider Caraway?
Very unlikely (1) → A little likely (2) → Somewhat likely (3) → Quite likely (4) → Very likely (5)
| Format | Very unlikely | A little likely | Somewhat likely | Quite likely | Very likely | Avg |
|---|---|---|---|---|---|---|
| Letterboxed | 9% | 23% | 27% | 19% | 23% | 3.24 |
| Side Blur | 16% | 22% | 17% | 20% | 26% | 3.18 |
| Outpainted | 10% | 15% | 24% | 24% | 27% | 3.44 |
Takeaway: Outpainted drives the most purchase consideration: a 51% top-two-box ("quite" or "very likely") against 42-46% for the alternatives. The framing treatment moves the needle on the metric closest to revenue.
Based on this video, how would you rate Caraway as a company?
Poor (1) → Fair (2) → Good (3) → Very good (4) → Excellent (5)
| Format | Poor | Fair | Good | Very good | Excellent | Avg |
|---|---|---|---|---|---|---|
| Letterboxed | 3% | 17% | 30% | 27% | 23% | 3.50 |
| Side Blur | 5% | 15% | 28% | 28% | 25% | 3.52 |
| Outpainted | 0% | 12% | 30% | 33% | 25% | 3.71 |
Takeaway: Outpainted earns the highest company rating, and the only perfect "zero poor" scorecard. Not a single viewer rated Caraway poorly. The frame doesn't just lift the ad; it lifts the brand behind it.
How would you describe this video's quality?
Looks amateur (1) → Somewhat amateur (2) → Mixed (3) → Professional (4) → Very professional (5)
| Format | Amateur | Somewhat amateur | Mixed | Professional | Very professional | Avg |
|---|---|---|---|---|---|---|
| Letterboxed | 9% | 17% | 37% | 31% | 5% | 3.07 |
| Side Blur | 8% | 14% | 39% | 26% | 14% | 3.24 |
| Outpainted | 4% | 8% | 33% | 39% | 15% | 3.55 Sig. |
Takeaway: Same footage, a higher-budget impression. 54% call the Outpainted version professional or very professional, versus 36% for Letterboxed, a gap large enough to be statistically significant (p = 0.001 vs. Letterboxed, p = 0.043 vs. Side Blur). Black bars read as amateur; a full frame reads as produced.
How much does this video look like it was made for a TV screen?
Not at all (1) → A little (2) → Somewhat (3) → Quite a bit (4) → Very (5)
| Format | Not at all | A little | Somewhat | Quite a bit | Very | Avg |
|---|---|---|---|---|---|---|
| Letterboxed | 23% | 17% | 22% | 23% | 16% | 2.93 |
| Side Blur | 27% | 21% | 15% | 16% | 22% | 2.85 |
| Outpainted | 11% | 11% | 21% | 29% | 29% | 3.54 Sig. |
Takeaway: This is the whole game. 58% say the Outpainted version looks made for TV versus about 38% for the others, a 0.6-point jump that is highly significant against both (p = 0.002 each). Letterboxed and side-blur are statistically tied and both feel like phone video pasted onto a television. Outpainting is the only treatment that clears the "is this even a TV ad?" bar.
Grouped
by concept.
Group scores combine the 1-5 metrics within each concept. Best in each row is bolded; Sig. marks a statistically significant lead.
| Concept | Metric | Letterboxed | Side Blur | Outpainted |
|---|---|---|---|---|
| Warm-up | Q1 enjoyment (1-5) | 3.28 | 3.46 | 3.43 |
| Attention / skip-resistance | Q3 attention (1-5) | 3.10 | 3.24 | 3.35 |
| Q2 watch-score (1-3) | 2.28 | 2.25 | 2.35 | |
| Purchase intent | Q4 + Q5 avg (1-5) | 3.11 | 3.15 | 3.34 |
| Brand perception | Q6 company rating (1-5) | 3.50 | 3.52 | 3.71 |
| Production value | Q8 quality (1-5) | 3.07 | 3.24 | 3.55 Sig. |
| Platform fit | Q9 made-for-TV (1-5) | 2.93 | 2.85 | 3.54 Sig. |
Overall composite.
Mean of all seven 1-5 metrics.
| Format | Composite (1-5) | vs. Letterboxed |
|---|---|---|
| Letterboxed | 3.16 | baseline |
| Side Blur | 3.23 | +0.07 |
| Outpainted | 3.46 | +0.30 |
The bottom line.
Outpainting isn't marginally better. It's categorically the right answer for UGC on CTV.
- It wins where it counts, with proof. On the two questions that define the format problem (does this look made for TV, and does this look professional), outpainting is the only treatment to win, and it wins by statistically significant margins. Letterboxing and side-blur are statistically tied with each other and both fail the "looks like a real TV ad" test.
- The quality halo carries downstream. Outpainting leads on attention, skip-resistance, brand rating, and both purchase-intent measures. Each lean is individually directional, but all five fall the same way, and they compound into a +0.30 composite lead over letterboxing. That's not noise; that's a pattern.
- It's the same footage. No reshoot, no new creative, no added spend, just a smarter way to fill the frame. The entire lift comes from the framing decision alone, which makes outpainting one of the cheapest quality upgrades available in a CTV buy.
- The default is costing you. Letterboxing, the path of least resistance, finished last or tied-last on every single metric. Every black-bar impression is leaving attention, brand favorability, and consideration on the table.
Recommendation: make outpainting the standard for UGC to CTV, and retire letterboxing as a default. To lock in the purchase-intent story with statistical certainty, re-run at about 200+ per cell. But the direction is already unambiguous: fill the frame, and the frame works for you.
Methodology footnotes: all percentages are within-format (column %). 1-5 means computed on respondents who answered each item. Significance via two-sided Mann-Whitney U. Two survey items from the original instrument (a brand-attribute multi-select and a brand-recall check) were excluded from this report due to noisy, non-discriminating data.
Want this result
for your creative?
Send us a vertical clip and the screens you need. We'll return a full-frame cut built from your original footage. No crop, no stretch, no reshoot.