reply to post by maxxsee
First off, I want you to know I have a very extensive background in computer graphics.
I want to tell you that you are resizing your images completely wrong, there is multiple ways to do it, but only ONE way to do it for analysis. When
you resize an image for analysis purposes you are supposed to use "pixel resize". That will preserve all the original pixels and colors, and only
multiply the size of them.
What you are doing is using "smart resize" or a similar function such as "bicubic" or "bilinear" which uses a not-so-smart algorithm to recreate
the original image. What it does is dilate the colors of the image to make them larger, but that also makes them blurry and changes the original
pixels, and it loses data.
So here is the correct way verses the wrong way:
You see the difference? The first picture preserved the EXACT PIXELS that your camera created. The second picture is a computer generated version of
the pixels from your camera. The first one is lossless, and the second one is lossy.
Although the second one looks smoother and better, it is only designed to "fix" images to look more visually pleasing after being resized, and it is
NOT designed for analysis purposes.
Here is a further resize of the actual pixels from the camera:
If you actually do an RGB color sample of each pixel, you can see that the flagpole is actually in FRONT of the umbrella.
You can see the flagpole RGB color actually runs through the umbrella, and you can see a single line of pixels that is darker than the umbrella on top
of the umbrella. However because of distance, and the width of the flagpole, the color of the umbrella and the pole bleed together slightly. This
happens because the pole is not exactly 1 pixel wide, and not exactly 3 pixels wide, it is about 2 pixels wide however it is trying to be displayed in
the center of 1 pixel. That means it would have to use 1 and two half pixels on each side, but you can't use half a pixel... so it bleeds.
The pole is in front of the umbrella.
Take note that I also save all my pictures in the ".PNG" format because PNG is lossless compression, unlike JPG which is lossly.
Next, we analyse the building:
I see three windows with white trim on a brown exterior wall on the upper floor. Right below that is a brown door, next to white poles, with a white
exterior on the lower floor. To the right is another deck with a white tarp liner. To the left is the flag pole, next to the umbrella....
It all lines up.... 100%.
It is an umbrella! There is nothing more to debate.
[edit on 24-11-2009 by ALLis0NE]