Use cases
Worked examples on Sume's catalog Formats and endpoints: product video and photography, try-on, app and product ads, logo animation, and more.
Start with Ready-made Formats for product video: the Sume Format catalog
Animated or fantastical AI videos: YouTube's disclosure exemption
YouTube says non-realistic content such as fantastical scenes or fully animated video needs no AI disclosure. Music as the main focus still does.
Live transcription for Korean meetings: MAI-Transcribe-2-Streaming
MAI-Transcribe-2-Streaming claims 60 languages with auto detection and about 100ms to first text. Verify Korean yourself. Sume transcribes files, not live.
Amazon will flag AI-generated people to shoppers: what to plan
Amazon says it will add a shopper indicator where an image includes AI-generated people, in all stores. Here is what is known, what is open, and how to plan.
Suno v6 on licensed data: does it change brand commercial-use risk
Suno v6 is reported to use licensed training data, but suits continue and commercial rights still need a paid plan. What changes for a brand and what does not.
Suno retires older models: keeping a brand jingle reproducible
TechCrunch says Suno will retire older models after v6. A brand jingle needs a stored master file and a saved prompt, not a promise to regenerate it.
Place the EU AI icon where app UI and captions will not cover it
The EU icon page says no overlay should obstruct the icon on video. TikTok publishes ad safe-zone files; use compose layout and caption placement to stay clear.
EU AI icon on video: burn it in, do not rely on a platform badge
EU icon guidance says the label must be embedded in the video and kept on reshare or download. A burned-in overlay does that; a platform badge does not.
TikTok Shop bans AI-enhanced before-and-after clips: claims checklist
TikTok Shop bars digitally enhanced before-and-after, instant-result claims and medical claims. Lint caption text before you burn it into a video.
TikTok says good video may lift views 50-70%: test it per SKU
TikTok's 2026 guide says good video may help increase views by up to 50-70%. Treat it as a claim: run two Format variants per SKU and compare your numbers.
TikTok Shop wants motion: no slideshows or screenshots
TikTok Shop's guides say avoid still-frame content and require 3 s of dynamic content per video; LIVEs ban over 50% still frames. Generate motion clips.
TikTok Shop price claims: burn the conditions next to the price
TikTok Shop wants price conditions disclosed: time limits, purchase minimums, location limits. Burn them as caption cues beside the price, $0.20 up to 60 s.
Keep a label intact while swapping a scene: anchor box or mask
BFL's FLUX 3 anchor box preserves a region; its doc example reports 85.5% pixel-identical unmodified regions. On Sume, use a ChatGPT Image 2.5 mask_url edit.
FLUX 3 Image free editing demo: a 4-edit test list before you pay
Test recolor, move, replace and swap in the reported free FLUX 3 precise-editing demo, then carry the winners to an API. Sume has no FLUX 3.
How Pinterest applies the AI label to a generated image Pin
Pinterest labels an image Pin when the owner says it is AI-made or its detection systems think so. Appeal steps are in a separate Help Center article.
Pinterest video ads: 15 minutes allowed, 6-15 seconds recommended
Pinterest allows video ads from 4 s to 15 min and up to 2 GB but recommends 6-15 s. Most Sume video models top out at 15 s, so one generation fits the band.
Replace headline text on a finished ad with Ideogram 4.5 on Sume
Send the finished ad as the first reference to ideogram/ideogram-v4.5 on Sume, name the new headline, omit aspect_ratio. List cost $0.03-$0.22.
Qwen-Image 2.1 circle-and-edit prompts, 10 references: a test plan
Qwen-Image 2.1 lists circle annotations, mask edits, 10 references. A test plan, plus the Sume route: mask_url and 16 references on ChatGPT Image 2.5.
Midjourney 100K-image history crash fixed; archive your files anyway
Midjourney's Oct 2 changelog fixes a crash opening images in 100K+ histories. Keep your own copy: download outputs and log prompts, settings and cost.
Midjourney collaboration is 'upcoming'; hand off with Sume metadata
Midjourney's Oct 2 changelog lists collaborative generation and edit history as upcoming. On Sume, put a handoff note in the job metadata field.
Seedream 5.0 Flash style blending: a five-step brand board test plan
A five-step plan to test multi-reference style blending on Seedream 5.0 Flash, with a Sume seedream-5-lite request to run the same test on a catalog model.
Landscaping yard transformation clip: Seedance 2.5 first+last frame
Send the muddy yard as first_frame and the finished lawn as last_frame to Seedance 2.5 on Sume: 8 seconds at 16:9 is $2.15 at 480p, $4.62 at 720p.
X carousel ads: 2 to 6 media assets, rendered as one Sume batch
X carousel cards hold 2 to 6 media assets. Make the whole set at one aspect ratio in a single Sume Image API call with n, then check each file.
LinkedIn Lead Gen Form image 552x200: make it from a 3:1 render
LinkedIn recommends a 552px by 200px form image. Sume has no 552:200 ratio, so ask Ideogram for 3:1, center-crop about 8% of the width, and resize.
Pinterest carousel ad specs: 1:1 or 2:3, 2 to 5 images, 20 MB
Pinterest carousel ads take 1:1 or 2:3 images, 2 to 5 cards, 20 MB each. Render a matching set in one Sume Image API call with n and one aspect_ratio.
X image ad specs: 1:1 or 1.91:1, 800px wide, 3 MB, from Sume
X website image cards take 1:1 or 1.91:1, at least 800px wide and 3 MB. Render 2:1 on Sume, crop to 1.91:1 with Pillow, and check size before upload.
Talking-head ad: Omni audio, Kling audio on, or Avatar Video?
Pick by length and face. Avatar Video runs 4 to 60 s from a script; Omni makes 3 to 10 s with native audio; Kling 3 adds audio at $0.21 a second.
Check prices survive translation before burning live-selling captions
A live-selling caption with a wrong price costs a $0.20 re-burn and a customer. Compare the digits of each source and translated cue with a ten-line check.
Person holding your product: reference clip caps on Omni, Wan and H3
To put a person with your product in one clip, send a photo plus a short reference clip. Omni takes 3 clips of 3 s, Wan 5 clips and 15 s, H3 3 clips and 15 s.
Barbershop before-and-after clip: Kling 3 first frame and last frame
Send the before photo as first_frame and the after photo as last_frame to kling-3 on Sume: a 6-second silent clip is $0.84; what to check before posting.
LinkedIn video ad intro text 150 characters, headline 70: test copy
LinkedIn video ads show 150 characters of intro text and a 70-character headline. Test copy on one rendered clip and save Sume jobs for hook variants.
Pet adoption listing clip: MiniMax H3 at 768p for 6 seconds
A shelter dog photo becomes a 6-second clip with minimax-h3 on Sume for 45 cents at 768p; stereo audio is always on, so there is no mute switch.
X post media mix: 4 photos, 1 GIF or 1 video, and 0.5 s video minimum
An X post can carry up to 4 photos, or 1 animated GIF, or 1 video, and video must be at least 0.5 seconds. Pick the right Sume export for each case.
Florist bouquet photo to a 5-second Reel clip: Wan 3.0 at 480p vs 720p
Animate a bouquet photo into a 5-second 9:16 clip with Wan 3.0 on Sume for 31 cents at 480p or 63 cents at 720p; a 12-bouquet weekly set costs $3.75 or $7.50.
Sales demo no-show follow-up: a 20-second AI avatar video and its cost
A no-show follow-up clip from one Sume avatar: a three-beat script, 9:16 or 16:9, the API call, and the cost for 50 missed demos on standard, plus and max.
X lead generation form fields: work email, job title, company
X's lead generation docs list contact and user-information fields, including WORK_EMAIL, JOB_TITLE and COMPANY_NAME. Pick fields to match each ad variant.
Instagram Feed video ad primary text: 125 characters, 30 hashtags
Meta's Instagram Feed video ad page lists primary text at 125 characters and 30 hashtags at most. Check bulk-generated copy against both before upload.
LinkedIn Lead Gen Form hidden fields: tag each lead with its variant
A LinkedIn Lead Gen Form allows up to 20 hidden fields. Keep your own index-to-variant table from the Sume bulk queue so every lead traces to one creative.
X lead generation form: name up to 255 characters, purpose 1000
X's Ads API lead generation docs cap a form name at 1 to 255 characters and its purpose at 1 to 1000. Validate copy variants before you create the form.
Pinterest CSV publish date: UTC ISO format to schedule Sume clips
Pinterest bulk CSV publish dates use UTC like 2023-12-17T08:00:00, and a blank cell publishes now. Generate a week of schedule slots for finished Sume clips.
Photographer portfolio reel: four portraits, four Seedance Fast clips
Four portraits, four 4-second 720p clips with seedance-2-fast on Sume cost $4.84; join them on a timeline for one reel.
Pinterest ad URL parameters: four UTMs applied automatically
Pinterest can add utm_source, utm_medium, utm_campaign and utm_content for you at account, campaign or Performance+ ad group level. Name variants to match.
Pinterest URL parameters do not override your feed or ad UTMs
Pinterest's automatic URL parameters will not override ones already in a product feed or a destination URL field. Clean your links before a variant test.
Call center roleplay training: customer persona clips with avatars
Record angry or confused customer prompts for agent practice with Sume avatars: one avatar per video, 10-second scripts, cost of a 24-clip scenario bank.
Open house invitation clip: MiniMax H3 Max, 8 seconds at 768p
A listing photo to an 8-second 16:9 open-house invitation with minimax-h3-max on Sume costs $0.80 at 768p, $0.50 at 480p and $1.60 at 1080p.
AI avatar video reply to a support ticket: 20-second script and cost
Send a short talking-avatar answer instead of a wall of text: one Sume avatar, a 20-second script per ticket, the request, and what 100 replies cost by tier.
Threads 250 posts in 24 hours: budget your Sume clips per day
A Threads profile can publish 250 posts in 24 hours, and a carousel counts as one. Work out how many Sume clips fit under that cap with a short script.
Which scripts can Sume burn into captions? Latin and Hangul, then test
Index-Translate covers 150 languages, but Sume's caption docs describe Latin and Hangul styles. Route by script and spend $0.20 on a test cue for the rest.
Sermon series title loop: Wan 3.0 first and last frame from one image
Make a looping sermon-series title background: send one image as both first and last frame to Wan 3.0 on Sume; 10 seconds costs $1.25 at 720p.