: Low vs high pixel dimenion downsampling If i have two images and one image is 7360 × 4912 and the other 3888 × 2592 and i downscale them both to 3000 px width - does either of the image

@Cofer715

As @Ken said, there will be a difference, here what I've noticed trying to downscale different pictures using different scaling algorithms.


Cubic: Preserves quality, it's used in a lot of image processing software, you will have likely the same result resizing those images to 3000px but you will need to sharpen the large picture a little bit if there is a lot of contrast differences.
Linear: This algorithm can't handle pictures with a lot of contrast irregularities, better not trying it.
Sinc (Lanczos3): It provides sharpening and preserves the quality as well, you can give it a try.

Vote Up Vote Down

@Barnes313

Yes, there will be difference. The larger image need to have more pixels averaged than the smaller one which means in effect that it may appear more blurry than the smaller one. This has to do with how interpolation works when down-sampling an image; it is basically functioning as a low-pass filter.

Another factor is what size of the resampling area is used for each pixel. If bi-linear then 2x2 pixels are considered while 4x4 are considered when using bi-cubic.

If the downscale step is too large you will get more aliasing when using bi-linear compared to bi-cubic. You can get around this by down-scaling in several steps.

But as the sizes you're using seem to be within at least what bi-cubic can handle I would suggest only giving the larger image a light sharpening after it has been down-scaled.

Vote Up Vote Down