I understand (I believe) that when I select a resolution smaller than my native, pixels are dropped from each frame. I'm wondering how much of a burden this is on my CPU.

I have a 1440p monitor, options to downscale resolution in Dxtory (recording gameplay), and options to downscale resolution in Blender when rendering. Would it be best for me to downscale when writing to file from Dxtory? Or perhaps, to downscale when rendering in Blender? Or (doubtful), would it simply be less hassle to write to file and render in 1440p?

Essentially, if I'm only going to do one (which I assume is the best method), when should I do it?

Thank you.


2 Answers 2


Blender's VSE always scales the image, unless the source is the same size as project.

The project's render properties determine the speed of the scale process. In the properties panel look for Anti-Aliasing.

Render Properties anti-aliasing

For best speed of scale turn this off. However the image will have jagged edges.

No anti-aliasing

To improve quality of the VSE scaling or other transform effects, you must turn on the Anti-aliasing. There are many levels and algorithms to choose from. The render-speed change may not seem to great, but you must multiply the difference out over the total number of frames rendered. And this example was based on a very simple image.

Anit-aliasing VSE


There are multiple overlapping concerns.

My philosophy is to conserve things at their "original resolution" for as long as is reasonable in the work flow. If I'm playing a video game at 2650x1440 then I would want the recording to be at 2560x1440 (also, where did hypothetical me get such a sweet monitor? Real me is derping along with a 1080p) . I would then include that video into my blender VSE and let blender do the downscale while it is compositing in overlays and doing cuts and wipes. Compositing video is probably the EASIEST thing I ever ask my computer to do (compared to rendering videos where a single frame is a 5-minute task, or computing an OpenCL mandelbrot that takes 17 minutes per frame).

A conflicting concern is: can your computer handle capturing the action at 1440 scan lines? If it can not keep up, it's probably computationally easier to downscale before video compression, even if you do lose resolution. I once tried to record an Overwatch session on a 5-year-old PC and it had multi-second gaps that ruined the recording.

So: record at native resolution and downsample at the latest stage in the workflow that makes sense, unless real-time concerns force you to downsample earlier.

  • $\begingroup$ Thank you for your input! I'll see how well I can record at 1440p. I know it's not that relevant, but could you explain what "composting video" is? $\endgroup$
    – Tawm
    Jul 15, 2016 at 20:36
  • $\begingroup$ I use the term "compositing video" to refer to layering multiple VSE strips. Some common reasons to do this are 1) adding titles or other text/graphical overlays 2) scene transitions (wipes/fades) 3) compositing in CGI effects (explosions, Jar Jar, visible "laser" beams) 4) concatenating various scenes into a longer video. $\endgroup$
    – Mutant Bob
    Jul 18, 2016 at 15:21

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .