That vore has a suspicious quantity of legs... 

there's an extra leg?

also q1CompilingGui source: https://github.com/necros0/ne_q1CompilingGui

i wrote this code ages ago... try not to laugh too hard. :P 

hail buttery satan 

http://www.dinofarmgames.com/a-pixel-artist-renounces-pixel-art/

This article is about the author's reasons to move away from pixel art to modern high-def graphics. There are some interesting thoughts about game art with numerous examples.

The comments are worth reading too. The author doesn't mention the visual differences between old and modern display hardware, but someone brings this point up. 

Every singleplayer Quake map played in a continuous marathon-like live stream (by multiple people), excluding the speedmaps (which are reserved for Daz!!). For extra exposure/attention it could even feature donations for some charitable cause. Kind of like the SDA events. OR all played by Daz and the donations used for the then-necessary psychologist sessions - which, of course, would be live-streamed as well! 

yupp http://www.quakeexpo.com/forum/viewtopic.php?id=5 

 

http://rickyt23.com/files/gimp-downsampling-looks-like-ass.html

Or am I missing something?

To me I get jaggies and artifacts from the fence on all images except image 5. 

Nice job picking the perfect ugly edge case for resampling. Patterns like that are hard to downsample. I guess PS detects them and blurs before resampling. 

Aren't "Sinc" and bicubic supposed to be the same thing? Guess Photoshop has some funky stuff going on in addition to the regular bicubic resampling to get better results. 

Photoshop does it correctly.

GIMP does it incorrectly.

My understanding is that if you reduced the size of an image like that by a factor of 4 then 16 pixels will become 1 pixel. The colour of that pixel *should* be dictated by the median colour of all 16.

What I see happening in GIMP is that this does not happen. What I see happening in Photoshop is that this does actually happen.

I mean....

Well here is the original image anyway:

http://wilmingtonfencecontractors.com/wp-content/uploads/2012/06/aluminum_fence_4.jpg 

Well, give it a try. Write that algorithm you described and see what happens. ;) 

Well I have oversimplified bicubic downsampling algorithm, there is another factor to the equation regarding the pixels furthest from the centre having less providence.

But you are also missing something. I mean look at the tree in the background. Everyone looks at the fence, but look at the tree! 

the gimp results are bad. Gimp also has trouble with test patterns like this: http://www.komar.org/faq/camera/auto-focus-test/micro-auto-focus-test-2.gif

I don't have photoshop, but Mac OS X's Preview app can resize that down with no artifacts appearing.

I do love free software and all, and on the bright side, someone motivated can fix it and send in a patch :-/ 

*prominence 

I downloaded the original image and ran a resize operation with the convert tool from ImageMagick. This was the command line:

convert aluminum_fence_4.jpg -resize 300x225 fenceout.jpg

And this is the result:

http://i.imgur.com/wJXnSD3.jpg

It looks similar to the good result from Photoshop to me. Thus, I don't think there are any fancy content-aware algorithms at work here.

I'd recommend adding ImageMagick to your toolbox.

http://imagemagick.com/script/index.php 

playing with it a bit, found this guide which recommends the following:

convert ~/Downloads/aluminum_fence_4.jpg -colorspace RGB -resize 300x225 -colorspace sRGB fenceout.png

result. This is better than photoshop I think? Note the left side of the fence doesn't go to a black blur as much. 

Hm, I thought moire artifacts were normal to expect from resampling if you did not add blurring beforehand.

Checkout graphicsmagick, there rarely is a reason to prefer imagemagick to it. It is much faster and stable.