Screen capture video output with lower quality than on-screen despite high res recording

[drumek]

Hi everybody,

I am recording my screen and camera picture in the corner (online course).
My laptop screen resolution is 1366x768.
Camera (Logi Brio 4k) set to record in 1920x1080.
Canvas 1920x1080
Scaled 1920x1080

Output video is indeed 1920x1080 but the screen recorded part looks way worse than originally on my laptop screen 1366x768.

Hope I made it clear.

Can you please let me know how to fix it?

https://obsproject.com/logs/gWN2I3FbKhnd6iZn

 

[koala]

Yes, it's no wonder why the video looks worse than your original screen. You tell OBS to upscale 1366x768 to 1920x1080, this is bloating up the video and making it blurry. If you view this 1920x1080 video on your 1366x768 monitor, it's downscaled again by the media player, making it even more blurry.

To have a clean 1920x1080 video, the source you're recording fullscreen must be 1920x1080 as well or above. If your laptop supports only 1366x768, connect an external 1920x1080 monitor, run your apps there and record this monitor.

If you don't have a 1920x1080 monitor, record your video with1280x720. This way you don't have an upscaled video. May be it will look even better if you set your desktop resolution to 1280x720 as well during recording. This will give you 1:1 pixels in the video as on your screen.

 

[drumek]

Thank you very much, thit explains a lot.
I tried connecting my TV as external monitor and recorded. This seems to work better.
Do I get it right that, when recording in higher resolution and downscaling to 1080p it should not be a problem and the video would look sharp as if it was actually recorded in 1080p? To make it happen does the initial resolution have to be 4K to downscale to 1080p or it can be anything between 1080p and 4K, I tried it this way but while rendering (in Movavi) the cropped part looked blurred.

Thank you!

 

[koala]

Any kind of scaling is making the video somewhat blurry, because with downscaling multiple pixels need to be combined to one pixel, or with upscaling multiple pixels need to be extrapolated to create a new pixel. So if you want a 1080p video, your source should be 1080p. This results in exactly the captured pixels.
Downscaling is required if you cannot have your high resolution source in the same (lower) resolution as your video. Upscaling is to be avoided if possible, because it adds no picture information but bloats the video. If you have a too small source, it's better to record the smaller resolution to the video and let the media player upscale on the fly during playback - in this case, there are no compression artefacts from the upscaling process.

Using a 4k source downscaled to 1080p is not sharper than original 1080p - it will also have some blurriness due to the downscaling algorithm. You get the sharpest image, if the original pixels are exactly 1:1 the same pixels as in the video, so source size = video size.

However, what you should do depends on the actual video content. Recording windows desktop apps with much text and thin lines, the above is true. Recording movie-like footage, outdoor with plants and such (doesn't matter if real or computer generated), will not show as much blur during rescaling, so you need to look yourself how much rescaling affects the quality of your video.

 

[drumek]

Thank you very much. This is really helpful.

Let me clarify one specific case. I am recording myself facing camera (just me in the center).
I tend to make mistakes while talking and in the editing I need to cut out mistakes and glue correct items together. Sometimes it looks bad because my head position (or hand gestures) is different on those parts that I want to come together. To avoid that I thought that I could zoom in second part to make it look like it is shot from different camera.

Would it make sense to record in 4k and have extra pixels to leave out during editing? Mix wide angle and close-up?

This should let me crop to 1/4 of original footage from 4k fram if my output video is to be 1080p.

Is it possible to have both wide angles (4k or slightly cropped) and close-ups (digital zoom in) have of the same quality in output video?
Currently close-ups are blurred even though I crop by less than 1:4 ratio.

I would appreciate your guidance. Thank you.

 

[konsolenritter]

Cropping 1:4 is HUGE! Even cropping 1:1.5 will do the trick. Recording a front camera in 4K is well overshoot, personally i think.

Leaving the OBS world possibly a transition effect in Davinci Resolve may help in your situation:

How To Use Smooth Cut Transitions In Davinci Resolve - Understanding Art Basics

I wanted to encourage others to cultivate their love of art. Read more on this website to find out how to create great art in less time.

tintedwave.com

But that's out of the obs support focus. You may try for yourself.

 

[koala]

Yes, it makes sense to record a higher resolution video, so you can crop part of it in postprocessing without quality loss to avoid upscaling the cropped part.

But you need to actually record this high resolution. If you record your desktop with whatever you want to present and want to include a camera video of yourself in some corner or cut the video and switch from the desktop to a fullscreen video of yourself and back, you need to record a full resolution video of your camera in addition to the full resolution video of your desktop.

So if you have your desktop with 1366x768 and your camera with 4k, and you want to composite this yourself with a postprocessing tool, you need to record both videos separately. One with 1366x768 and the other with 4k. Resizing, cropping and placing both over each other (this is called compositing) can be done in a postprocessing tool ("Video editor").

However, your laptop is too weak to make such 2 recordings simultaneously. Its even not powerful enough to record something with 4k resolution.

Some people do this: they double the frame size and put their main source next to the camera source.
For example, they want their final video be 1280x720, with a desktop recording and a camera of theirselves.
They double the canvas and output size to 1280x1440. In the upper half of that canvas, they put their desktop source with resolution 1280x720 and in the lower half of that canvas they put their camera source with also 1280x720. This can be used to record both simultaneously with OBS.
After recording, they use a video editor to separate both halves and have a 1280x720 camera video of theirselves to put wherever they like into the other half, the desktop recording.

 

[drumek]

Yes, it makes sense to record a higher resolution video, so you can crop part of it in postprocessing without quality loss to avoid upscaling the cropped part.

But you need to actually record this high resolution. If you record your desktop with whatever you want to present and want to include a camera video of yourself in some corner or cut the video and switch from the desktop to a fullscreen video of yourself and back, you need to record a full resolution video of your camera in addition to the full resolution video of your desktop.

So if you have your desktop with 1366x768 and your camera with 4k, and you want to composite this yourself with a postprocessing tool, you need to record both videos separately. One with 1366x768 and the other with 4k. Resizing, cropping and placing both over each other (this is called compositing) can be done in a postprocessing tool ("Video editor").

However, your laptop is too weak to make such 2 recordings simultaneously. Its even not powerful enough to record something with 4k resolution.

Some people do this: they double the frame size and put their main source next to the camera source.
For example, they want their final video be 1280x720, with a desktop recording and a camera of theirselves.
They double the canvas and output size to 1280x1440. In the upper half of that canvas, they put their desktop source with resolution 1280x720 and in the lower half of that canvas they put their camera source with also 1280x720. This can be used to record both simultaneously with OBS.
After recording, they use a video editor to separate both halves and have a 1280x720 camera video of theirselves to put wherever they like into the other half, the desktop recording.

Thank you, this is very helpful. I want to crop only an image of myself (then make transition to screencast+myself), so that in the video it is only myself. I tried recording in 1080p then crop it slightly (really - just cut out the roughly 10% off from each of the edges) and render output video in 720p. The cropped part looked worse than the rest. I don't know, can this be related to video editing software I am using (Movavi)?

 

[koala]

I'm unable to help with video editing, sorry. This is something everyone has to work out for himself. There is so much video editing software available, from trivial to stuff used by Hollywood, from free to multi-thousand Euro.

 

[Jean de Suiza]

What's the best choice for the average user and novice editor?

Camtasia when you want an easy solution with a lot of utilities to produce an impressive result without having to learn a lot...
Otherwise, I recommend PremierePro, it's very complete and professional, specially when you work with shortcuts.
Both solutions are relatively cheap in comparison with others.

 

[koala]

Start with the video editor that comes with Windows 10. You find it in the start menu by searching for "video editor".