Resizing Images – Computerphile


Nearest Neighbour and BiLinear resize explained by Dr Mike Pound

Fire Pong:
Google Deep Dream:
FPS & Digital Video:

This video was filmed and edited by Sean Riley.

Computer Science at the University of Nottingham:

Computerphile is a sister project to Brady Haran’s Numberphile. More at



  1. I was attracted by the title 'resizing images' expecting a load of bollocks – and I was right.
    We're offered two "solutions" – make the pixels bigger – which doesn't alter the size of the image; OR create a new image from the original – which also doesn't alter the size of the original image.

    If you want my advice about changing the size of an image, if you want to make it smaller, stand further away and if you want to make it bigger, get closer to it.

  2. I think my best analogy to this is like driving on the wrong side of the road. You can do it but that doesn't mean it's right.

  3. Nearest neighbor is kind of like , take the pixel and make multiple cells "equal" to one pixel. Zooming in, so to speak. Just pretend that one pixel cell is now a 2×2 pixel of cells, in other words. So for example you could scale by twos simply and quickly without any math by just converting every pixel into a 2×2 grid of pixels.

  4. I was kind of disappointed that you didn't show an example of how a scaled image with bilinear (and bicubic) interpolation would look, preferably the same image that was used as an example of nearest-neighbor. Other than that, great video as always!

  5. Can someone please explain downscaling an image that is high res? Even with more data to work with, the results aren't always that great ?

  6. I wonder how would it in nearest neighbor if the new pixel is exactly in the middle. Basically happens when scaling ratio is an even number.

  7. There is program, now that uses deep learning to increase artificially the resolution of a picture. It's an interpretation from an AI, so, it's not perfect.

  8. Actually, if you're a pixel artist its a bad thing to use that form of nearest neighbor because you want to keep pixel aspect ratio. Good pixel artists will use what Sean says, where 1 pixel becomes 4, 9, 16, etc. I think the more practical form of nearest neighbor (and better illustration) is where 1 pixel becomes the center of a larger grid of pixels depending on your scaling ratio.

  9. I don't understand what you say about scaling down… if I scale down there's no more "pixel values that sit in between the original pixels". I have to throw away some of my original pixels, and map them differently. Maybe if it's a x2 down sizing, take an average of every 4 pixels in the original image to represent 1 pixel in the down-sized image.

  10. Could you use deep learning to intelligently upscale and interpolate images, alongside changing exposure and other settings?


Please enter your comment!
Please enter your name here