The media boxes INPUT, PREVIOUS, NEXT, and REFERENCE allow you to select which videos or images are processed in the next operation.
There are four ways to select media:
- Click a media box once to select it, then click it again to open a file chooser dialog to import a new media asset.
- Drag media from a file browser window into a media box. It will be imported into Color Correct.
- Copy media to the clipboard, select a media box, and press
⌘VorCTRLV. The new media asset will be imported. - Select a media box, then click the
+button on any existing media gallery item to add that item to the media box.
Using a Single Frame Pair and a Previous Video Clip
Use this mode to color-correct a video clip that has been generated using the end frame of a previous clip.
- Add the video clip you want to color-correct to
INPUTand the preceding video clip toPREVIOUS. - Ensure the single frame pair option is selected in the command palette.
- Drag the green frame markers over matching frames from each video. Typically the last frame of
PREVIOUSand the first frame ofINPUTwill match. - Click the green Go Button. The color correction profile will be built from the matching frame pair and applied to all frames of
INPUT.
Using a Single Frame Pair and a Next Video Clip
Use this mode to color-correct a video clip that has been generated using an end frame that matches the start frame of another clip.
- Add the video clip you want to color-correct to
INPUTand the successive video clip toNEXT. - Ensure the single frame pair option is selected in the command palette.
- Drag the blue frame markers over matching frames from each video. Typically the last frame of
INPUTand the first frame ofNEXTwill match. - Click the green Go Button. The color correction profile will be built from the matching frame pair and applied to all frames of
INPUT.
Using a Double Frame Pair
Use this mode to color-correct a video clip that has been generated using the start and end frames of a reference clip. Note: this is only necessary if the reference clip itself contains a shifting color profile; otherwise you can match using a single frame pair.
- Add the video clip you want to color-correct to
INPUTand the reference video clip toPREVIOUS(NEXTcan be used instead for the same result). - Ensure the double frame pair option is selected in the command palette.
- Drag the green frame markers over matching frames from each video. Typically the first and last frames of
PREVIOUSand the first and last frames ofINPUTwill match, respectively. - Click the green Go Button. The color correction profile will be built from the matching frame pairs and applied to all frames of
INPUT, interpolating the color correction between the first and last frame pairs for a smooth color profile transition.
Using Color Balance Heuristics
Use this mode to color-correct a video clip to match the color balance of a reference image using video color heuristics rather than frame-based pixel matching. Use this mode only when you do not have a least one pair of matching frames.
- Add the video clip you want to color-correct to
INPUTand the reference video clip toPREVIOUS(NEXTcan be used instead for the same result). - Ensure the general color match option is selected in the command palette.
- Click the green Go Button. The color correction profile will be built using the heuristics of
PREVIOUSand applied to all frames ofINPUT.
Use this mode to color-correct a video clip to match the end frame of a previous clip and the start frame of a next clip.
- Add the video clip you want to color-correct to
INPUTand the preceding and successive video clips toPREVIOUSandNEXT, respectively. - Ensure the double frame pair option is selected in the command palette.
- Drag the green frame markers over matching frames from
PREVIOUSandINPUT. Drag the blue frame markers over matching frames fromINPUTandNEXT. Typically the last frame ofPREVIOUSand the first frame ofINPUTwill match as well as the last frame ofINPUTand the first frame ofNEXT. - Click the green Go Button. The color correction profile will be built from the matching frame pairs and applied to all frames of
INPUT, interpolating between the first and last frame pairs.
Use this mode to color-correct a video that was generated with a raw RGB start frame that did not pre-compensate for color space conversion drift, or when a video-to-video process results in the same type of color drift.
- Add the video clip you want to color-correct to
INPUT. - Ensure that the Color Recovery option is selected in the command palette.
- Set the Color Recovery level. For videos created with image-to-video AI, try level
2. For videos created with video-to-video AI, try1. - Click the green Go Button. Color Recovery will be applied to all frames at the specified level.
Use this mode to correct the internal color drift that can occur in an AI-generated video.
- Add the video clip you want to color-correct to
INPUT. - Ensure that the Drift Correction option is selected in the command palette.
- Move the green marker to the reference video frame with the color balance that you want apply to the rest of the video – usually the first frame.
- Move the blue marker to the video frame with the greatest amount of drift – usually the last frame.
- Click the green Go Button. The color drift within the clip will be reduced or eliminated.
- Note that this operation works best when the two specified frames have similar content and lighting, such as the same characters in the same scene. Depending on the variance between the frames
Use this mode to export a video frame and pre-compensate for the color space conversion drift that will occur in an AI video generated using that frame.
- Add the video clip you want to color-correct to
INPUT. - Ensure that the Extract Image option is selected in the command palette.
- Move the green marker to the frame you want to export (default: last frame).
- Set the Color Recovery level. In most cases this will be
2. - Click the green Go Button. The specified frame will be exported with the specified level of Color Recovery applied. An AI video generated using this start frame and exported Color Recovery level
2will automatically have the correct color balance. - Typical Color Recovery levels:
- Exports the raw RGB without modification.
- For frames that will be directly turned into a video by Color Correct or a similar app.
- For frames that will be used as AI video start frames or end frames.
- Available if necessary.
Use Rescale to fix small size, scale, and position mismatches between an AI-processed output and the original — most often a Seedance clip that visually matches its start frame but appears slightly stretched or shifted. Rescale recovers the horizontal and vertical scale (up to ±2% per axis) and translation that best aligns INPUT to the reference, then resamples INPUT to the reference's dimensions.
Rescaling a Video
Use this mode to resize an INPUT video so its horizontal and vertical size, scale, and position align to a reference video or image. Useful for correcting an AI-processed output that does not match the original size and/or scale and may appear slightly distorted.
- Add the video clip you want to rescale to
INPUTand the reference clip or image to eitherPREVIOUSorNEXT. - Ensure the Rescale option is selected in the command palette.
- Drag the green frame markers over matching frames in
INPUTand the reference. Typically the last frame ofPREVIOUSand the first frame ofINPUTwill match. If the reference is an image, only theINPUTframe marker matters. - Click the green Go Button. The recovered scale and translation will be applied to every frame of
INPUT, and the output will be written at the reference's dimensions.
Rescaling an Image
Use this mode to resize an INPUT image so its horizontal and vertical size, scale, and position align to a REFERENCE image. The two images should have similar content. Useful for correcting an AI-processed output that does not match the original size and/or scale and may appear slightly distorted.
- Add the image you want to rescale to
INPUTand the reference image toREFERENCE. - Ensure the Rescale option is selected in the command palette.
- Click the green Go Button. The
INPUTimage will be resampled to the dimensions ofREFERENCEunder the recovered scale and translation.
Use Lip-Sync to sync a character's mouth movements to audio using the sync-3 model. It works on a video (syncing to the video's own audio track) or on a still image paired with an audio file.
Lip-Sync a Video
Use this mode to lip-sync a character in a video to its audio track using the sync-3 model.
- Add the video you want to lip-sync to
INPUT. It must contain an audio track. - Ensure that the Audio Operation tab is selected in the command palette and that the Lip-Sync button (the leftmost button on the upper option bar) is selected.
- Adjust the green and blue frame markers to specify the range to apply the lip-sync to. By default they cover the full clip. For best results, include a few frames before and after spoken audio.
- If more than one person appears in the video, click on the
INPUTmedia to open the media view, then click and drag to define a selection rectangle around the speaker's face. The selection rectangle should surround the speaker's face across all frames specified by the frame markers but does not need to be exact - too large is better than too small. - Click the green Go Button. A new video will be created with the speaker's mouth movements synced to the audio track over the selected frame range. In addition to the usual base credit cost and size cost multiplier, the lip-sync operation costs 10 credits per selected frame.
Lip-Sync an Image (Talking Avatar)
Use this mode to turn a still image and an audio file into a video of the character in the image speaking the audio.
- Add the audio file and the image in any order to
INPUTandMEDIA. - Select the Lip-Sync button (the rightmost button on the option bar) in the command palette.
- Set the desired frame rate. Most AI-generated video is 24 frames per second.
- Click the green Go Button. A new video will be created of the character speaking the audio at the selected frame rate. The video is as long as the audio file, and the speaker is detected automatically (unlike the video mode, image lip-sync does not support a speaker selection rectangle). In addition to the usual base credit cost and size cost multiplier, the lip-sync operation costs 10 credits per frame, counted as the number of 24 FPS frames needed to fully contain the audio (the frame-rate selection does not change the cost).
Use Overlay to composite a second image or video on top of INPUT, pinned to a corner, edge, or the center. OVERLAY is layered without resizing — pixels falling outside INPUT's frame are clipped, and transparent regions of OVERLAY reveal INPUT underneath. Handy for watermarks, lower thirds, logos, or stacking a generated insert onto a base shot.
- Add the base clip or image to
INPUT, then add the watermark, logo, or insert to eitherPREVIOUSorNEXT. - Select the Overlay option in the command palette.
PREVIOUSmedia is automatically moved to theNEXTslot and the slot is relabeledOVERLAY; thePREVIOUSslot is replaced by theANCHORpicker. - Click an
ANCHORcell to pinOVERLAYto a corner, edge, or the center ofINPUT.ANCHOR - Drag the frame markers on
INPUTto select the interval (inclusive) during whichOVERLAYshould appear. Defaults to the full clip.- If
OVERLAYis an image, it shows only during that interval. - If
OVERLAYis a video, its first frame is aligned to the first marker. If the overlay is shorter than the interval it finishes early; if longer it is clamped at the second marker.
- If
- Click the green Go Button. The output keeps
INPUT's dimensions, format, and frame rate;OVERLAYis composited at its native size.
Use this mode to clip a portion of a video.
- Add the video you want to clip to
INPUT. - Ensure that the Edit Operation tab is selected in the command palette and that the Clip button (the second button on the upper option bar) is selected.
- Move the green marker to the first frame of the new clip and the blue marker to the last frame of the new clip.
- Click the green Go Button. A new clip will be created that includes the range of frames specified by the two markers.
Use this mode to crop a video to a rectangular region, and optionally to a specific range of frames.
- Add the video you want to crop to
INPUT. - Ensure that the Edit Operation tab is selected in the command palette and that the Crop button (the leftmost button on the upper option bar) is selected.
- Click on the
INPUTmedia to open the media view, then click and drag on the video to define the cropping rectangle. Only one rectangle is allowed at a time — drawing a new one replaces the previous, and clicking an existing rectangle removes it. - Drag the green and blue frame markers on
INPUTto define the range of frames to keep (inclusive). Defaults to the full clip. - Click the green Go Button. A new video will be created that contains only the pixels inside the cropping rectangle, across the selected range of frames.
Note: if you ensure that the cropped region is flush with at least one corner, you can modify the resulting clip with a lipsync tool or other tool, color-correct the result of that if need be, and then use the Video Overlay operation to easily reintegrate the modified clip into the original clip.
Use Reframe to conform a video or image to a standard aspect ratio and size — for example, turning a slightly off-size generated clip into a clean 16:9 1080p video, or reframing a landscape still as a 9:16 vertical. The media is scaled to cover the target frame (preserving its aspect ratio, upscaling if necessary) and then cropped to the chosen anchor — the result is never letterboxed. The output is resized to the nearest standard size (720, 1080, or 2160 on its smaller dimension) and named with a -Reframe suffix. If the media already matches the selected aspect ratio at a standard size, the Go Button is disabled because no reframe is needed.
Video Reframe
- Add the video you want to reframe to
INPUT. - Ensure that the Edit Operation tab is selected in the command palette and that the Reframe button (the third button on the upper option bar) is selected.Then select the target aspect ratio. The aspect ratio nearest the source is selected automatically when you enter Reframe.
- Use the
ANCHORwidget (shown in place of thePREVIOUSinput) to choose where any cropping is taken from — for example, anchor to the top to keep heads in frame when narrowing a landscape clip to vertical. The default is center.ANCHOR - Click the green Go Button. The video is resized and cropped to exactly match the selected aspect ratio at a standard size.
Image Reframe
- Add the image you want to reframe to
INPUT. - Ensure that the Reframe tab is selected in the command palette and select the target aspect ratio. The aspect ratio nearest the source is selected automatically when you enter Reframe.
- Use the
ANCHORwidget (shown in place of thePREVIOUSinput) to choose where any cropping is taken from. The default is center.ANCHOR - Click the green Go Button. The image is resized and cropped to exactly match the selected aspect ratio at a standard size.
Use this mode to splice (join) two or three video clips into a single video.
- Add the video you want to splice to
INPUT,PREVIOUS, and/orNEXTmedia boxes. - Ensure that the Splice option is selected in the command palette.
- For each selected media, move the green marker to the first frame that should be spliced into the new clip and the blue marker to the last frame.
- Click the green Go Button. A new clip will be created that includes the selected frames from the two or three selected videos.
Using Pixel Matching
Use this mode to color-correct an input image based on a reference image using a pixel-matching algorithm. This works best when the two images have matching content.
- Add the image you want to color-correct to
INPUTand the reference image toREFERENCE. - Ensure that the Pixel Match option is selected in the command palette.
- Click the green Go Button. The Input image will be color-matched to the Reference image using pixel matching.
Using Color Balance Heuristics
Use this mode to color-correct an input image based on a reference image using color heuristics.
- Add the image you want to color-correct to
INPUTand the reference image toREFERENCE. - Ensure that the Color Balance Match option is selected in the command palette.
- Click the green Go Button. The Input image will be color-matched to the Reference image using image heuristics.
Use this mode to split a clip's audio into two separate MP3 stems: a Voices track containing just the isolated voices, and a Sounds track containing everything else (music, ambience, and effects). The separation is performed by an AI model and runs on the entire clip. You can extract stems from either a video or a standalone audio file. When the source is a video, a third file is also created — a muted copy of the video with a silent audio track — which is shown as the result.
- Add the video whose audio you want to separate to
INPUT. - Ensure that the Audio Operation tab is selected in the command palette and that the Extract Stems button (second from the left on the upper option bar) is selected.
- Click the green Go Button. Three new files will be created — one for the isolated Voices, one for the remaining Sounds, and a muted copy of the video with a silent audio track (shown as the result).
When INPUT holds an audio file on its own, the command palette offers Extract Stems directly — there is no Audio Operation tab to select.
- Add the audio file whose stems you want to separate to
INPUT, leavingPREVIOUSandNEXTempty. - Click the green Go Button. Two new MP3 files will be created — one for the isolated Voices and one for the remaining Sounds. (There is no muted video output for an audio-only source.)
Use this mode to extract a span of the audio track from a video to a standalone MP3 file. By default both frame markers cover the entire clip, which extracts the full audio.
- Add the video whose audio you want to extract to
INPUT. - Ensure that the Audio Operation tab is selected in the command palette and that the Extract Audio button (the rightmost button on the upper option bar) is selected.
- Drag the green and blue frame markers on
INPUTto narrow the range. Both markers are inclusive — the extracted audio matches the length of the span they define. Leave the markers at the first and last frame to extract the entire audio track. - Click the green Go Button. A new MP3 audio file will be created containing the audio between the two frame markers.
Use this mode to mute part — or all — of a video's audio track without otherwise changing the video. The output keeps the input's dimensions, frame rate, length, and video stream; only the audio between the two frame markers is silenced.
This pairs naturally with Lip-Sync: when two characters speak in rapid succession, silencing the other speaker's audio first lets sync-3 lock cleanly onto the target speaker's voice instead of being pulled between two voice sources.
- Add the video whose audio you want to silence to
INPUT. - Ensure that the Audio Operation tab is selected in the command palette and that the Silence Audio button (the rightmost button on the upper option bar) is selected.
- Drag the green and blue frame markers on
INPUTto define the span to silence. Both markers are inclusive. Leave them at the first and last frame to silence the entire audio track. - Click the green Go Button. A new video file will be created with the audio between the markers silenced; audio outside that range is preserved at its original level.
Use this mode to attach an audio track to a video. Any existing audio in the video is replaced. The result is the same length as the source video; audio that's longer than the video is truncated, audio that's shorter leaves the trailing portion of the video silent.
- Add the audio file to
INPUTand the video toMEDIA. The command palette will automatically switch to audio+video mode. - Ensure the Add Audio button (left tab on the option bar) is selected.
- Click the green Go Button. A new video will be created using the source video's frames and the audio file as its soundtrack.
Use this mode to mix an audio track on top of a video's existing audio so both play together — for example, layering music or narration over a clip without losing its original sound. Unlike Add Audio (which replaces the audio), Overlay Audio keeps the original track and blends the new one in at full volume. If the video has no audio of its own, the new track is simply added. The result is the same length as the source video; audio longer than the video is truncated.
- Add the audio file to
INPUTand the video toMEDIA. The command palette will automatically switch to audio+video mode. - Select the Overlay Audio button (right tab on the option bar).
- Click the green Go Button. A new video will be created with the audio file mixed over the source video's original soundtrack.
Use this mode to color-correct a frame extracted from a video that color-drifted from the original start frame.
- Add the image you want to color-correct to
INPUT. - Ensure that the Color Recovery option is selected in the command palette.
- Set the Color Recovery level. In most cases this will be
2. - Click the green Go Button. Color Recovery will be applied to the image at the specified level.
Use this mode to erase the background of an image. The output is a PNG with transparency where the background used to be.
- Add the image whose background you want to remove to
INPUT. - Ensure that the Remove Background option is selected in the command palette.
- Ensure that the AI removal method (the default) is selected in the Removal Method bar that appears above the command palette. The second button switches to the Chroma Key method.
- Click the green Go Button. The background will be removed and the result saved as a transparent PNG.
Chroma Key a Video
Use this mode to key out a video's dominant backdrop color frame by frame. The key color is detected automatically per frame, and the transparency is delivered in the output format of your choice.
- Add the video you want to chroma-key to
INPUT. - Ensure that the Edit Operation tab is selected in the command palette and that the Chroma Key button (the rightmost button on the upper option bar) is selected.
- Choose an output format in the Output Format bar that appears above the Edit Operation bar.
Mask(the default) produces two MP4 videos: a color video with the backdrop's contribution removed from edge pixels (suffix-Deblend) and a grayscale alpha-mask video (suffix-Mask, listed as the older of the two). Use this pair with compositing tools that accept separate fill and matte inputs.ProResproduces a single ProRes 4444.movwith the alpha channel embedded — the professional interchange choice, but the files are very large (see the ProRes note under How Are Credit Costs Calculated?).VP9produces a single VP9.webmwith embedded alpha — compact and web-friendly, though Safari cannot play these files (the gallery shows the preview thumbnail with a video-file icon instead; the file can still be downloaded). The ProRes and VP9 formats are available on paid plans only. TheMaskandVP9outputs can also be combined or converted into a ProRes file on your own computer — see Convert Chroma Key Outputs to ProRes Locally. - Click the
INPUTvideo to open the media view. It shows a live preview of the chroma key result for the current frame over a checkered transparency backdrop — hover along the bottom of the frame to scrub and check other frames. - Optionally click and drag on the preview to draw exclusion masks that protect areas from the key. Protected pixels always stay fully opaque across every frame, and they don't contribute to the automatic key color detection.
- Optionally adjust the
ToleranceandSoftnesswidgets that appear in place of theNEXTmedia box. Drag the green slider or click the value field and type a new number — the live preview updates as you tune.Tolerance(0.00–0.40, default 0.03) sets how close a pixel's color must be to the key color to become fully transparent — raise it if a noisy backdrop leaves speckles behind.Softness(0.00–2.00, default 1.00) scales the width of the transparency falloff at the subject's edges: lower values give harder edges, higher values keep more of the semi-transparent fringe. - Optionally set a
Hair Colorif the subject has wispy hair (or similar fine detail) at its edges. TheHair Colorbox appears in place of thePREVIOUSmedia box while Chroma Key mode is active. Click the box, then click and drag over a sample of hair in the media view — the average of the sampled colors becomes the repair color, shown as a solid swatch in the box (hover the swatch and click itsxto clear it). Very translucent edge pixels whose color lies between the backdrop and the sampled color are pulled toward it during deblending on every frame, restoring the hue and opacity the backdrop had swallowed; differently-colored edges are detected and left untouched, so no halo is added elsewhere.Hair Color - Click the green Go Button. Every frame is keyed and the output (or output pair) is rendered in the selected format. The thumbnail shows the first frame with its transparency applied.
Chroma Key an Image
Use this mode to key out an image's dominant backdrop color (for example a green screen). The key color is detected automatically and the output is a PNG with transparency where the backdrop used to be.
- Add the image you want to chroma-key to
INPUT. - Ensure that the Remove Background option is selected in the command palette.
- Select the Chroma Key method in the Removal Method bar that appears above the command palette.
- Click the
INPUTimage to open the media view. It shows a live preview of the chroma key result over a checkered transparency backdrop. - Optionally click and drag on the preview to draw exclusion masks that protect areas from the key. Protected pixels always stay fully opaque, even when they match the key color, and they don't contribute to the automatic key color detection. Protect any part of the subject that shares a color with the backdrop.
- Optionally adjust the
ToleranceandSoftnesswidgets that appear in place of theNEXTmedia box — the same controls as for videos, but the image tolerance defaults to 0.01. - Optionally set a
Hair Colorfor subjects with wispy hair — the same control described for videos above, appearing in place of theREFERENCEmedia box. Click the box, click and drag over a sample of hair in the media view, and the averaged color repairs the translucent hair edges. - Click the green Go Button. The chroma key will be applied and the result saved as a transparent PNG.
Without Audio
Use this mode to turn a single frame into a freeze-frame video. The duration of the output video is 1 second.
- Add the image you want to turn into a video to
INPUT. - Ensure that the Create Video option is selected in the command palette.
- Set the desired frame rate. Most AI-generated video is 24 frames per second.
- Click the green Go Button. The image will be converted into a freeze-frame video.
With Audio
Use this mode to combine a still image and an audio track into a freeze-frame video. The output video's duration matches the audio track's duration.
- Add the audio file to
INPUTand the image you want to freeze on toMEDIA. The command palette will automatically switch to Create Video mode. - Set the desired frame rate. Most AI-generated video is 24 frames per second.
- Click the green Go Button. A freeze-frame video will be created using the image as every frame and the audio file as its soundtrack.
Color Balance and Drift Correction operations use one of the algorithms listed below. Different algorithms work best for various cases. The other frame-pair-matching algorithms use a custom algorithm.
'%3e%3ccircle%20r='15'%3e%3c/circle%3e%3crect%20x='12'%20y='-5'%20width='24'%20height='10'%3e%3c/rect%3e%3crect%20x='12'%20y='-5'%20width='24'%20height='10'%20transform='rotate(45)'%3e%3c/rect%3e%3crect%20x='12'%20y='-5'%20width='24'%20height='10'%20transform='rotate(90)'%3e%3c/rect%3e%3crect%20x='12'%20y='-5'%20width='24'%20height='10'%20transform='rotate(135)'%3e%3c/rect%3e%3crect%20x='12'%20y='-5'%20width='24'%20height='10'%20transform='rotate(180)'%3e%3c/rect%3e%3crect%20x='12'%20y='-5'%20width='24'%20height='10'%20transform='rotate(225)'%3e%3c/rect%3e%3crect%20x='12'%20y='-5'%20width='24'%20height='10'%20transform='rotate(270)'%3e%3c/rect%3e%3crect%20x='12'%20y='-5'%20width='24'%20height='10'%20transform='rotate(315)'%3e%3c/rect%3e%3c/g%3e%3c/g%3e%3c/svg%3e)