Confused about resolution vs. quality

[Track]

I'm creating a time-lapse and I have a general question about resolution vs. quality.

I have an 18MP photo.

I re-size it to 2MP (1920x1280) and view it at 100% - it looks super-sharp.
http://hcd-1.imgbox.com/aboQHasC.jpg?st=cUjEUNQHoWgAUc8GIaWQNw&e=1378067524

I re-size it to 8MP (3840x2560) and view it at 50% - it looks blurry.http://hcd-1.imgbox.com/abgV58bU.jpg?st=W64QLKsm26aDVXvalBupQQ&e=1378067517

If I view the latter at 100% and zoom into the former to match the length, they both look identical with the latter have a slight advantage is sharpness.

So, basically.. my question is - why is is that when you re-size a photo to a lower resolution is looks better than viewing the photo at a lower percentage? In the past, whenever I viewed a photo like that, it always looked sharper. I never realized I could enhance the sharpness further by re-sizing it..

Thanks!

[poisondeathray]

It depends on the resizing algorithm used.

e.g. if your software viewer uses a soft bilinear algorithm for resizing the 8MP to 50%, it's going to look soft compared to a lanczos or spline36

Same with the software and algorithm used to resize it to 2MP

[Track]

It depends on the resizing algorithm used.

e.g. if your software viewer uses a soft bilinear algorithm for resizing the 8MP to 50%, it's going to look soft compared to a lanczos or spline36

Same with the software and algorithm used to resize it to 2MP

So, what do I do?


I'm rendering a time-lapse and I need to know if re-sizing it to 3840x2560 will not have lower quality than 1920x1280.

[poisondeathray]

It depends on the resizing algorithm used.

e.g. if your software viewer uses a soft bilinear algorithm for resizing the 8MP to 50%, it's going to look soft compared to a lanczos or spline36

Same with the software and algorithm used to resize it to 2MP

So, what do I do?


I'm rendering a time-lapse and I need to know if re-sizing it to 3840x2560 will not have lower quality than 1920x1280.

So it depends on what format goal is, and how you intend to watch it, what resizer your viewing hardware/software is using, and what bitrates you intend to use

If you are viewing on a 1920x1080 display , it probably makes zero sense to resize to 3840x2560 for viewing purposes

But if you have adequate bitrate, the higher resolution version will look better on a 4K display

[budwzr]

There are three attributes for resolution. "Pixel Dimensions", "DPI", and "Inches". They are inversely proportional, and any two determine the third. So the "Best" attributes boil down to the target output. AND the resize algorithm as mentioned by my esteemed colleague.

Also, resizing can be accomplished by crop as well, thus avoiding any resampling.


================================
"I donn need no stinkin badge"

[Track]

I still don't understand.

When you're viewing an 8MP photo and a 2MP on a 4MP display, the software will show them at the same size meaning that it will show the 2MP at 100% and the 8MP at 25%. Showing it at 25% should be equivalent to re-sizing it to 2MP.. but in reality, it looks worse than if I were to actually go and re-size it myself to 2MP.


So, what is the culprit? Is it the software I used to re-size it, is it the software I use to view it.. or do images simply look better at 100%?

[budwzr]

If your monitor is 96ppi, and your image is 72ppi, your image would display smaller, but still be at native resolution. If your image is 96ppi, and your monitor is 72ppi, the image would display larger, and also be at native resolution.

When working with raster images, you can only see them properly at native resolution. Otherwise, you're seeing a resample. In the better image software, the zoom level will always be prominent so you can flip back to it as a reference and preview any changes.

When editing video you have to match the project to the source, and set the preview window size accordingly, to see the native resolution.

And turn off resampling, and whatever else might be auto-resizing the media. Or if you can't you have to use whatever tools available, like TrackMotion, or overcrop.

A 640X360 media clip, coming into a 1280X720 project should be 1/4 of the preview window to be in native resolution.

When resizing down there's a curious effect of perceived sharpness, and this is due to the contrast increase due to the reduction of intermediary pixels between colors. But actually, there's a reduction of detail and your mind fills in the lost detail from what it remembers of the original image.

Icons are an example of this. They look sharp because they're "concentrated" or "distilled". Like milk-water turns to cheese. Like a big puffy dough ball gets the poof knocked out. Some software even use the term "inflate" to upsample.

PDR already gave you all you need to know, but I sense that you don't always know where "1:1" is, or how to get to it, and how to use it as your reference. So I hope this helps, but feel free to reiterate where your confusion lies.

The more I expound, the more risk I will misstate something, due to brain fart, and I'll get hammered by one of those high flying B-2 stealth bombers constantly on patrol.