Converting a 1080p project to 2.7k

Is there a way to convert an existing 1080p project with video footage, filters, etc. to 2.7k? I tried changing the resolution of the existing project to 2704x1520 via the settings → video mode → custom, but when I play it back, things do not look quite right. Ultimately what I am trying to achieve is to make an export video for YoutTube in 2.7k resolution. If I just export my current 1080p project to a 2.7k output it does not fill the entire frame. BTW, if you are wondering why I want to export a 2.7k for YouTube is to force it to use its better codec. Any help with this would be appreciated.

Thank you. It looks like my best option at this point is to export the video to 1080p and then start a new project in 2.7k to resize and export again to produce my 2.7k.

A small hint - From the point of view of mathematics, there is no point in converting the picture with a smaller resolution into a picture with a large, because in the process of primary digitization, a significant part of the initial analog (continuous time) of information is caused by the discreteness of the countdown in time. The camera receives a continuous input in time the data stream and during operation performs measurements of its parameters n times per unit of time, but because For the operation of the equipment, we need to reset the measuring circuits before each new dimension, there are time intervals between the measurements during which we do not have information about the visible chamber of the scene - that is, part of the input information is lost (see the theorem of the Kotelnikov on the sampling of the irresistible input signal), and with the raising Forced to use an estimated picture formed based on the surrounding selected point of neighboring “average of …” method.

And the more around the surrounding selected neighboring points we use, the closer the point we are built to the lost source, but we will never be able to fully achieve the compliance with the “calculated - initial” match.

Plus, it should be borne in mind that digital cameras are usually used by JPEG compression that is poorly scalable due to the features of the algorithm and for the lack of visual distortions of the output picture (color artifacts) should not be used to scale more than 200%. If more scaling is required, it is necessary to use several cycles of zooming input images with an intermediate smoothing of color noise and contrast correction, white balance and sharpness. And this will greatly increase the processing time. I had to apply this technique for cases when you need to see some small objects removed from a long distance, but the accuracy of playback does not give such a technique, although it is understood that we see we can. This technique is suitable for examples of biologists when observing some kind animals of which we are forced to shoot from a long distance, but it is applicable only in specific cases, and that with great reservations …

Mathmatically this is correct, but on Youtube this doesn’t apply. Uploading the same video as 2.7k/4k instead of 1080p will force youtube to use higher bitrate and different codec with pretty massive results in quality. Keep in mind that even if you upload 1080p with a bitrate of 60mbps, youtube will transcode it to 1080p with something like 4-5mbps (probably using avc codec) and you lose all the quality. 1440p has around 12mbps and 4k 24mbps and uses vp9 or av1.

So if you have 1080p footage at 24mbps or higher it will look better in youtube if upscaled and uploaded as 4k compared to just uploading it in 1080p.

2 Likes

Thank you for restating the purpose of the original topic. I wish Shotcut had an option to do this automatically when exporting for YouTube, i.e., to make a 2.7k version of the 1080p footage. Right now I have to make an intermediate 1080p output since my projects are in 1080p mode, and then import it back into a new 2.7k project to make the final video for YT. :disappointed:

I never tried this before but I have an idea for a quick test project with a one clip + one text overlay and it seems to work.

Try this: keep your project in the original 1080p resolution and only change the export panel - (click Advanced) and in the Video tab change the resolution to 2.7k (I gave a shot with footage of 1920x1080 and export of 2560x1440 and it seems to upscale everything properly without any black bars resulting in a 2560x1440 video clip). Note that this would be a quick upscale-like conversion, you need to start the project from the begining in 2.7k if you want original sharp edges on filter result.

2 Likes

Everything you said. Also use Lanczos as the interpolation algorithm on the Advanced export panel. There is also the option of adding a Sharpen filter on the Output track to make things even crisper. These are hacks, but they do help survive the upscale gracefully.

1 Like

Yep, thank you guys, this worked just fine. I will use it for my next YouTube video. I will send the link when I do. :smiley:

For those of you following this topic, I uploaded a couple 30sec video clips of the type of video I worked with (dirt bike riding) into YoutTube to demonstrate what a difference it makes when you force YouTube to use its better v09 codec by using 2.7k videos vs.1080p. The original video was captured with a GoPro camera set at 1080p, 59.94fps which match the video mode of the Shotcut project. One of the clips uploaded was exported at 1080p @59.94fps, and the second one was exported using a resolution of 2704x1530 also at 59.94fps. The two links are:

BTW, you can check which codec was used by right clicking on the video and selecting “Stats for nerds”. AVC1 for 1080p videos and vp09 for 2.7k/4k

4 Likes