Guides
Sume guides: the API, Formats, agents, models, media tools and pricing, each answer dated and sourced.
AI avatar for online course videos: build and update lessons
Use an AI avatar as your online course instructor: one reusable avatar, a short talking video per section, captions, and one Timeline join per lesson.
Lip sync AI for long videos: past the 5-minute clip limit
To lip sync a long video with AI, split the audio, sync each part to the same face, and join the clips. One Sume job takes up to 300 s of audio.
AI avatar for YouTube videos: Shorts and long-form
Use an AI avatar in YouTube videos: a 9:16 talking video of up to 60 seconds for a Short, or 16:9 segments joined into one long-form video.
How to burn an SRT file into a video
Burning an SRT file draws its lines into the picture. Sume's captions API takes no .srt file, so turn each block into a cue and send the cues instead.
How to generate an SRT file from a video
To generate an SRT file from a video, transcribe it with timestamps and write each timed sentence as a numbered block. Sume's STT returns timed JSON.
Hard subtitles vs soft subtitles: what's the difference?
Hard subtitles are burned into the picture and always show; soft subtitles are a separate track viewers can turn off. When to use each, and how.
How to add subtitles to a video from a transcript
Have the words but no timings? Send the transcript as script_text with the video to Sume's captions API, which times it to the speech and burns it.
How to add bilingual subtitles to a video
Burn two languages onto one video with Sume: put both lines in each cue of one caption job, or run two jobs, each language at its own height.
How to add a logo or watermark to a video
Add a logo or watermark to a video by laying the image over every frame. Sume's Timeline compose does it for Sume-hosted files in one MP4.
How to add an image to a video at a specific time
Show a picture only between two timestamps: compose it over that stretch of the clip, or cut to it as a still, then rebuild the video in one render.
How to convert MP3 to MP4 with an image
Turn an MP3 into an MP4 with one picture: a Timeline 1.0 render holds the image for the whole track, up to 30 minutes, at the frame size you set.
How to freeze frame a video: hold the last or any frame
Freeze a video's last frame with one tpad filter call, or hold any frame mid-clip as a still between two parts of the clip in a Timeline render.
How to increase video length by looping a short clip
Loop a short clip until it matches a song or voiceover: put it in one Timeline 1.0 slot as long as the audio and set render.pad_mode to loop.
How to remove silence from a video automatically
Find every pause from a transcript's word timings, keep the speech with a small margin, and render the kept ranges back to back in one Timeline job.
How to fix audio delay in a video by shifting the sound
Fix a video whose sound runs late or early: detach the audio and re-render the clip with the offset set as audio.source_in or the slot's source_in.
How to rotate a video 90 degrees and save it
Rotate a video 90 degrees with ffmpeg's transpose filter: clock turns it clockwise, cclock counterclockwise. Sume's video filter saves a new MP4.
How to make a video square: crop it or fit it in 1:1
Make a video square by cropping a centered 1:1 window, or fit the whole frame in a square with black bars. Both are one FFmpeg filter job on Sume.
How to make a vertical video horizontal (9:16 to 16:9)
To make a vertical video horizontal, put the 9:16 clip in a 16:9 frame and fill the sides with a blurred copy or black bars. How to do it with Sume.
How to remove black bars from a video by cropping
Black bars baked into a video only go away with a crop. Measure them on a full-size still, then crop them off with FFmpeg's crop filter on Sume.
How to make a video black and white (or keep one color)
Make a video black and white by setting its saturation to zero with FFmpeg's hue or eq filter, or keep one color with colorhold. How to do it on Sume.
How to brighten a dark video: exposure, gamma, or levels
Brighten a dark video with FFmpeg's exposure, eq gamma, or colorlevels filter, run as one Sume video filter job. To darken instead, use the dim op.
How to color correct a video: fix a color cast and warmth
Color correct a video with FFmpeg filters on Sume: colorbalance removes a yellow or green cast, colortemperature sets warmth, and eq sets saturation.
How to blur part of a video or black it out
Blur part of a video by cropping the area, blurring it, and overlaying it back in one FFmpeg filtergraph on Sume, or cover it with a solid drawbox.
How to merge two videos of different resolutions
Merge two videos of different resolutions by scaling both into one frame: crop, pad, blur, or stretch each clip, then join them with their sound.
How to add a voiceover to a video and keep its sound
Add a voiceover to a video by mixing narration over its own sound: speak the script with TTS, then render the clip with both tracks in one MP4.
MP4 video has no sound? Find where the audio went
An MP4 with no sound has no audio track, a silent one, or one your player can't decode. Check with a free probe, then fix the step that lost it.
How to convert HEVC (H.265) video to H.264
Convert HEVC (H.265) to H.264 by re-encoding the video stream. On Sume, an exact trim of a whole non-HDR clip returns an H.264 MP4.
How to reduce video file size to fit an upload limit
Reduce video file size with fewer bits: a smaller frame, a lower frame rate, a shorter cut, or no audio. Sume takes no bitrate, so check size_bytes.
How to crop a video without losing quality
A normal crop re-encodes the frames, so lose as little as you can: crop once, from the original, and don't scale back up. How Sume encodes each edit.
How to check a video's bitrate from its size and length
Check a video's average bitrate: size in bytes × 8 ÷ length in seconds. Sume's free probe returns both numbers, size_bytes and duration_seconds.
How to change the background of a photo with AI
Give an AI image model your photo and a prompt that names the new background, or cut the subject out first. How to do both with the Sume API.
How to get a white background product photo with AI
Cut the product out with a background remover and flatten the PNG onto white, or have an image model re-render the shot on white. Both with Sume.
AI lifestyle product photography from one packshot
Turn one packshot into lifestyle scenes: send it as a reference to an image model, describe one scene per request, and keep the shots that match.
AI mockup generator API: put your design on a product
Make product mockups with an image API: send your design and a blank product photo as references, say where the design goes, and check each result.
How to combine two photos into one with AI
Yes, AI can combine two photos into one: send both as references to an image model and describe the single scene you want. How to do it with Sume.
AI image aspect ratio changer: re-render at 9:16 or 16:9
To change an image's aspect ratio with AI, outpaint the new area or redraw the image at the new ratio. Sume has no uncrop tool; here is the redraw.
How to upscale an image to 3000x3000 (or any exact size)
Divide the target size by the current size to get the upscale factor, then run it: 1000×1000 at 3× is 3000×3000. Sume takes any factor from 1 to 4.
How to generate variations of an image with AI
Send the image as a reference, say what may change and what must stay, and set n for several results per call. How to do it on Sume's Image API.
Photo to painting AI: turn a photo into a painting
Turn a photo into a painting with AI: send it as a reference image, name the style in the prompt, and keep its shape with aspect_ratio auto.
Sketch to image AI: turn a drawing into a finished picture
Turn a sketch into an image with AI: send the drawing as a reference, describe the finished result, and ask for up to four versions per call.
Consistent character AI image generator: reuse one reference
Keep one character consistent across AI images: send the same reference image with every request, repeat its description, and change only the scene.
AI image seed: what it is and how to repeat an image
An AI image seed fixes a generation's random start so a result can repeat. Sume's image API does not serve seed yet; reuse a reference image instead.
AI image with text generator: get the words right
Make an AI image with text: quote the exact words in the prompt, raise the quality setting, proofread every letter, and set exact copy yourself.
AI image for print: pixel sizes for 300 DPI prints
A 300 DPI print needs 300 pixels per inch of print. Generate an AI image at the largest size a model allows, then upscale it up to 4x with Sume.
AI banner generator for YouTube: a 2560×1440 channel banner
YouTube recommends a 2560 x 1440 px channel banner, 16:9 and 6 MB or less. On Sume, ChatGPT Image 2.5 takes that exact size as a custom image_size.
AI banner generator for LinkedIn: a 1584×396 cover image
LinkedIn recommends a 1584 x 396 px cover image, a 4:1 strip. Generate it at 4:1 with Nano Banana 2 on Sume, then resize it to the exact size.
AI banner generator: wide banner shapes up to 8:1
Choose the banner's shape, then a model that lists it. On Sume, ChatGPT Image 2.5 takes exact sizes up to 3:1 and Nano Banana 2 goes to 8:1.
AI wallpaper generator: 4K desktop and phone wallpapers
Generate a wallpaper at your screen's pixel size. On Sume, ChatGPT Image 2.5 takes an exact 3840 × 2160 for 4K and near-exact phone sizes to crop.