Last week I tried to drop a short product demo into a Discord channel. The upload failed. Discord’s free plan caps a file at 20 MB, and a few minutes of screen capture went over that without much effort. I did not want to split the walkthrough into two messages, and I did not want to paste a cloud link that people might skip, so I looked for a video compressor I could run in the browser and then upload like any other file.
I record AI-tool walkthroughs often enough that this has become a regular chore: the clip looks fine on my machine, then the chat refuses the attachment.
The dead ends I hit first
My first try was to export again from the capture tool at a lower resolution. That can work, but I was guessing. One export was still too large. The next one squeezed the file down, and the labels on the interface turned soft. For a demo, those labels are the point.
The second idea was to cut the recording into two parts. Discord would accept each one, but the person watching had to play them in order, and the middle of the explanation sat in a second message. That is a poor way to show a single flow.
The third idea was a shared-drive link. I have sent those in chat before. Some people open them. Some people say they will watch later, and the later does not arrive. For a short preview, I wanted the file in the channel itself.
What I needed was narrower than a full edit: keep the same runtime, shrink the file, and aim at a size I could check before I hit upload.
Setting a target size in a video compressor
The setting that helped was not a quality slider I had to eyeball. It was a target size just under the 20 MB cap, with a little room left so the upload would not fail on a rounding difference.
I used a browser video compressor that accepts a file up to 10 GB, runs in the browser, and does not require an account. The output is an MP4, with no watermark. It takes the recording formats I actually have, including MP4, MOV, MKV, AVI, WebM, and WMV, plus a longer list beyond those.
The workflow I settled on looks like this:
- Save the screen recording as it is. Do not split it first. A large source file is fine, as long as each upload stays within 10 GB.
- Upload the file. Drag and drop works. If the recording already lives in Dropbox, or you have a direct file link, those import options are there too.
- Choose Basic mode and type a target size below the cap. For Discord’s 20 MB limit, I aim lower than 20 MB. The final size can vary slightly, so I treat the target as a ceiling I need to verify, not a promise.
- Start the compression and wait. How long it takes depends on the file size, the duration, the settings, the current server load, and my upload speed. The page shows progress while it runs.
- Check the output size before I download. If it still sits too close to 20 MB, I lower the target and run it again. The duration of the video stays the same.
- Download the MP4 and upload it to the channel. I only replace my original file if I am happy with the result and actually need the disk space.
If I need more control than a single target number, Advanced mode is there: quality, resolution, bitrate, and codec. H.264 is the safer choice when I do not know what the other person will use to play it. H.265 can be more efficient on devices that support it. I stay on Basic when the only requirement is “fit this in Discord.”
One more practical detail: the file is processed in the cloud, and result files are deleted within 24 hours. The recording is not used to train AI. I still avoid uploading anything I would not be comfortable sending through a normal file transfer.
What I watch for before I hit upload
A target that equals the cap is a trap. If the limit is 20 MB, a 20 MB file can still fail if the service counts a little overhead, or if the output lands slightly above the number I typed. I leave a margin and look at the actual output size.
Playback matters more than the file size alone. After I download the MP4, I scrub through the moments where I was talking over a dense UI. If the text on screen turns into a blur, I go back and raise the target a little, or I switch to Advanced and bump quality, then check the size again.
Format surprises are less of an issue than they used to be, because the download is an MP4. That is the container I want for chat. If the original recording came from a camera or an older capture tool in MOV or MKV, I do not convert it in a second app first. I upload the source and let the output come back as MP4.
I also do not treat this as a way to “improve” the picture. Compression can take out some detail. The aim is a file that still reads clearly and actually uploads.
If you need to compress a walkthrough to a target size for a 20 MB chat cap, Basic mode is the path I would start with. Type a number below the limit, check the output, and only then upload it. That last check is the step I used to skip, and it is the one that stopped the file from bouncing again.


Top comments (0)