For a 24/7 relaxation stream, set a deliberate output canvas—often 1920×1080 for a 16:9 landscape channel—and preserve each source video's display aspect ratio. If your clips have different shapes, FFmpeg can fit each one inside that canvas, but fitting alone does not fill the space left over; choose padding, a background or an intentional crop separately.
The filter examples below are adaptable patterns, not tested commands for every input, encoder or FFmpeg build. Check the source metadata, render a representative sample and validate the output before you rely on it for a live broadcast.
Choose the output canvas and inspect the source aspect ratio
Start with the frame you want YouTube to receive, not with a scale filter copied from another command. A 16:9 landscape canvas such as 1920×1080 is a common choice for a relaxation channel, but it is not mandatory. The right dimensions depend on your source material, the encoder you can run, available upload capacity and the quality you need. A smaller, clean source does not gain genuine detail merely because you enlarge it to 1080p.
Next, inspect the actual source. With FFprobe, request stream details and look for the coded width and height, sample aspect ratio (SAR), display aspect ratio (DAR) and pixel format. A source that is 1440×1080, for example, is not necessarily displayed as a 4:3 picture: its pixel aspect metadata may say otherwise. Do not infer the intended shape from width and height alone, particularly with older or anamorphic footage.
The distinction matters because the visible picture is governed by display aspect ratio. FFmpeg's FAQ on aspect ratio explains that scaling normally adjusts SAR to keep DAR constant, rather than stretching the image. If a clip looks unnaturally wide or narrow in a player, investigate its metadata and appearance before trying to compensate by changing dimensions. An incorrect DAR tag can make otherwise sensible scaling look wrong.
For a folder of relaxation visuals, sample more than the first file. Still images, phone footage, landscape loops and older archive clips may all have different proportions or metadata. Note the widest and tallest sources, whether any are interlaced, and whether their pixels are square. This small inventory helps you decide whether to fit the whole image, crop it, or place it against a designed background.
Fit variable sources inside a fixed box
For mixed aspect ratios, force_original_aspect_ratio=decrease tells FFmpeg's scale filter to fit the input within the requested width and height while preserving its proportions. A useful starting point is:
scale=1920:1080:force_original_aspect_ratio=decrease:force_divisible_by=2
This asks for dimensions no larger than 1920×1080. A 16:9 source can use the full box; a portrait or 4:3 source will be scaled until its width or height reaches the corresponding limit, leaving the other dimension below the limit. The result remains proportional. It is not necessarily 1920×1080 in its actual frame dimensions, and the filter by itself does not create pixels around the fitted picture to make a full-size canvas.
That distinction is easy to miss when you are making a stream from a playlist. One clip might occupy the full landscape frame, while the next is a narrower picture with unused space at the sides. If the output is meant to stay a uniform 1920×1080 frame through every transition, add a deliberate padding or compositing stage after scaling. If you want the original material to occupy more of the screen, consider an editorial crop only when losing its edges is acceptable.
The FFmpeg scale filter documentation describes the aspect-ratio options and their behaviour. Documentation evolves, and installed builds can differ, so check the documentation for the FFmpeg version you actually use. If you are building the rest of a command around a looping file, keep scaling as one part of the pipeline; audio mapping, looping, encoding and reconnection need their own checks. The FFmpeg bitrate checklist for a 24/7 stream covers the separate choice of ingest bitrate.
What force_divisible_by changes
force_divisible_by=2 constrains the dimensions produced by the aspect-ratio adjustment so that they are divisible by two. That is useful with common video encoders and pixel formats that expect even frame dimensions. It does not change the source's aspect ratio deliberately, fill empty areas or guarantee compatibility with every encoder setting.
There is a small practical compromise: an ideal proportional fit can produce an odd width or height. Requiring even dimensions means the filter must choose a nearby compatible result, so one dimension may be rounded to a suitable value. For a relaxation visual, that tiny dimensional adjustment is generally less problematic than a frame size rejected by the encoder, but you should inspect the rendered output rather than assume what a particular build will choose.
Use this option as a constraint, not as a quality setting. It does not improve detail or remove compression artefacts. If you work with unusual pixel formats, a hardware encoder or an older FFmpeg installation, check the filter output and the encoder's requirements together. The filter's documentation is the place to confirm the exact options supported by your local version.
Add padding or a background when the canvas must be filled
If the stream needs a fixed 16:9 output frame while preserving every part of a source, fit first and pad second. A centred pattern is:
-vf "scale=1920:1080:force_original_aspect_ratio=decrease:force_divisible_by=2,pad=1920:1080:(ow-iw)/2:(oh-ih)/2"
The pad stage adds canvas around the scaled picture until the frame reaches 1920×1080. The coordinate expressions centre the image horizontally and vertically. Unless you specify another colour or composite a background, the extra area uses the pad filter's configured background colour. It may be black, which can be perfectly acceptable for a dark ambience channel but may look like a letterbox around brighter footage.
For a more considered presentation, replace or cover the plain empty area with a suitable background. You might use a static colour that matches the visual, a blurred enlargement of the same clip, or a simple branded frame. The important point is that the background is an explicit design decision; the fit stage does not manufacture it. For a calming image of a candle in portrait orientation, for example, a muted blurred background can fill the landscape sides while leaving the candle itself undistorted and fully visible.
If you process a sequence of files, make the background and positioning consistent from clip to clip. A change from black bars to a bright background at every playlist boundary can be more distracting than the source's original aspect ratio. Test the transitions as well as the individual frames. When even empty margins are preferable to visual treatment, plain padding is simpler and preserves the full image.
Decide whether cropping is worth the loss
A filled frame can also be achieved by enlarging an image until it covers the canvas and cropping the excess. This is a different editorial choice from fitting and padding: it fills the frame by removing part of the source. A wide landscape clip may lose content at its sides when adapted to a taller shape; portrait material may lose content at the top or bottom when made to cover a wide frame, depending on the source and crop position.
Use crop-to-fill only when the image has room to lose those edges. A close view of a mountain ridge may tolerate a carefully chosen crop; a devotional image with text, a face near the edge, or a candle flame may not. Inspect important content and captions across the entire frame before adopting a crop. If you cannot accept that loss, preserve the full image and add padding or a designed background instead.
FFmpeg also offers force_original_aspect_ratio=increase for a scale operation that grows dimensions to cover a target box. It does not, by itself, make cropping disappear: a later crop stage is needed if the required output is to be exactly the box size. Treat an increase-and-crop arrangement as a deliberate composition, not as a substitute for understanding what parts of the image will be removed.
When simple 1920×1080 scaling is suitable
If you have verified that a source is already 16:9, square-pixel footage, a straightforward scale=1920:1080 may be enough to set the output dimensions. It is easier to read and avoids adding aspect-ratio options that you do not need for a known source. Confirm the displayed result first; a 16:9 width-to-height ratio in the file does not prove that its aspect metadata is correct.
Simple scaling is less suitable as a blanket rule for a folder of varied sources. A portrait clip forced to 1920×1080 without aspect-ratio preservation becomes wide and flattened. A 4:3 image forced to the same dimensions becomes unnaturally broad. Do not use this form to make non-16:9 footage fill the frame. Choose a fit-and-pad or crop approach based on whether preserving the whole picture or using the whole canvas matters more for that scene.
Resolution is also a source and delivery decision, not just a number to place in a filter. Upscaling a small file may make it fit the frame dimensions without supplying new texture or detail. A 720p output can be a sensible choice if your source is modest or your encoding and upload constraints make 1080p impractical. For a related look at the whole live pipeline, see how to choose an output bitrate for FFmpeg and YouTube ingest.
YouTube's encoder settings guidance ties recommended ingest bitrate to codec, resolution and frame rate. For example, its current guidance lists H.264 at 1080p30 at 14 Mbps, 720p30 at 8 Mbps and 480p30 at 4 Mbps; these are recommendations for those specific combinations, not universal quality guarantees. The same page covers other codecs, frame rates and advanced settings, so check it for current guidance rather than carrying over a bitrate from a different setup. Keep resolution, frame rate, codec and the capacity of your connection in view together.
Preserve display aspect ratio and avoid stretching
The safe default is to preserve DAR, the shape of the displayed picture. FFmpeg's FAQ summarises the mechanics: “Most processing filters in FFmpeg handle the aspect ratio to avoid stretching the image: cropping adjusts the DAR to keep the SAR constant, scaling adjusts the SAR to keep the DAR constant.” In plain terms, scale changes pixel dimensions while adjusting sample-aspect information where needed so that the visible image retains its shape.
This is why a fitted source may have non-square sample aspect metadata even if your intended output is square-pixel video. Check both the resulting dimensions and its reported display aspect ratio. YouTube's recommended live advanced settings include square pixels and progressive scan for typical output, but a filter chain should be tested with your actual input and encoder. Do not change SAR or DAR blindly just to make a player display a preferred shape.
If the source metadata is wrong, correcting the metadata can be appropriate, but only after comparing it with the actual image. FFmpeg's FAQ discusses setsar and setdar for intentional aspect-ratio changes. These are not universal repair buttons: setting a new ratio when the source is merely unusual can make the distortion worse. Keep an untouched copy and test a corrected output before replacing the source in a live playlist.
A good visual check is straightforward. Compare a recognisable circular object or a face in the source with the render; circles should remain circular and faces should not become noticeably broader or narrower. Also inspect any edge details that cropping could remove. A technically valid file can still be a poor composition for a channel, particularly when text or a devotional subject sits close to the border.
Test the rendered output before using it live
Render a short sample containing the kinds of sources that will actually appear in the channel: a verified 16:9 clip, a tall or square visual, and any older footage with unusual metadata. Use the intended filter chain and encoder. Inspect the output dimensions, SAR, DAR and pixel format with FFprobe, then watch it in a player that respects aspect metadata. Look for distortion, unexpected black borders, clipped edges, colour changes and transitions that shift the framing.
Keep a copy of the exact filter and command used for the successful sample. FFmpeg option behaviour can depend on version, and a filter pattern does not validate audio handling, looping, encoder availability or YouTube ingest settings. If you change the FFmpeg build, pixel format, encoder or source set, repeat the render. A small test catches a wrong crop or unexpected canvas before it repeats through a long broadcast.
Then make a private or otherwise controlled preflight stream if your workflow allows it. YouTube recommends testing with representative audio and motion, checking stream health and messages while live, and using a speed test to assess the connection. Its service can transcode a live stream into formats for viewers on different devices and networks, but that does not correct a malformed source composition or guarantee a reliable 24/7 process. Review current YouTube live encoder guidance before launch.
Scaling solves frame geometry; it does not keep an encoder running after a power cut, internet failure or process crash. If the machine must remain on to run FFmpeg, consider the practical cost of keeping it powered, connected and monitored. A cloud-based workflow can remove the need to leave your own computer running: StreamNeo turns an uploaded file into a YouTube live stream and restarts the broadcast if it drops, which addresses the specific burden of maintaining the local playback machine, not source rights or the design choices discussed here. For the broader trade-offs, read how to choose a cloud streaming service for 24/7 YouTube Live and, if you plan to keep a local encoder, how to check network settings when an OBS stream disconnects in India.
Before committing, compare the operating options on the pricing page. When the file and channel are ready, start free — 24-hour trial, no card.
FAQ
Does force_original_aspect_ratio=decrease fill a 16:9 frame?
No. It fits the source inside the requested box while preserving its proportions, so one dimension can remain smaller than the box. Add a pad stage or another deliberate background treatment if you need a full-size canvas without cropping the source.
Can I use scale=1920:1080 for every relaxation clip?
Use it directly only when the source's displayed aspect ratio is verified as 16:9 and its metadata is sensible. For a portrait, square or 4:3 visual, that direct scaling distorts the image; fit and pad it, or choose a considered crop instead.
What does force_divisible_by=2 do?
It constrains the dimensions selected by the aspect-ratio scaling behaviour to be divisible by two, which is useful for common encoder and pixel-format requirements. It does not fill canvas space or improve image detail, and the exact result should be checked with your installed FFmpeg build.
Will the right scale filter make a 24/7 stream reliable?
No. Scaling controls geometry, not power, internet stability, process recovery, YouTube ingest health or content rights. Test the rendered file and stream pipeline separately, then monitor the live stream and check current official guidance.