Useful IQ comparison tool...

[Dave Baumann]

If you can get over the idea that this is an ATI tool, ATI do actually have a very useful app for easily comparing image quality IQ differences. Take a look here:

http://www.ati.com/developer/tools.html

and download "The Compressonator".

Although this tools primary use if for compressed texture generation it also has an image quality compre tool. If you go to "File" -> "Compare Images..." (or "Compre Directories") you can open up two images. Once they are open there will be 3 displays - the first selected image on the left, the second on the right and in the middle an image difference window. if you move around or zoom the image in a window all follow suite and you can also increase the brightness on the image difference to highlight if there are any differences.

 

[micron]

How long have you known about this?

 

[Reverend]

If you can get over the idea that this is an ATI tool <snip>

Unfortunately, I can't and really shouldn't. I'm sure ATI will say this tool doesn't do any funky stuff (tool says "Hmm... an NV card detected... let's change things a bit...") but I always try to lean on the side of neutrality

 

[Pete]

So name all your images "nV*," rather than "ATi*" and "nV*."

 

[croc_mak]

If you can get over the idea that this is an ATI tool <snip>

Unfortunately, I can't and really shouldn't. I'm sure ATI will say this tool doesn't do any funky stuff (tool says "Hmm... an NV card detected... let's change things a bit...") but I always try to lean on the side of neutrality

You should download it and try...This is purely a image compression and comparision tool..It has nothing to do with any HW.

You can run this tool on S3 hardware and compare Nvidia or ATI images if you want to..This is just a 2d tool.

You should really feel stupid for the above comment

 

[micron]

You should really feel stupid for the above comment

Um nooo....noooo he shouldnt

 

[zurich]

But that tool assumes that the reference, 'correct' image is an ATI rendered one?

 

[Doomtrooper]

Interesting

 

[kyleb]

But that tool assumes that the reference, 'correct' image is an ATI rendered one?

i didn't download it but i don't see how the program would ever have any clue as to what video hardware rendered any images you may care to compare.

 

[K.I.L.E.R]

Dave, there are other programs that do the exact same thing. I found one a couple of years ago.

Unfortunately I don't have it now.

 

[Doomtrooper]

But that tool assumes that the reference, 'correct' image is an ATI rendered one?

i didn't download it but i don't see how the program would ever have any clue as to what video hardware rendered any images you may care to compare.

Exactly

 

[K.I.L.E.R]

It actually works quite well this program.

Unfortunately you need identical locations to take the screenshot from else you will be met with a worrisome difference.

 

[Dave Baumann]

Unfortunately, I can't and really shouldn't. I'm sure ATI will say this tool doesn't do any funky stuff (tool says "Hmm... an NV card detected... let's change things a bit...") but I always try to lean on the side of neutrality

Rev, download it and take a look.

The tool was designed for comparing compressed with uncompressed textures. You read in two images (it copes with numerous image formats) an then you can analyse them - it can't know where those images have come from or what has rendered them, a, say, .bmp doesn't embed any information! You wouldn't want any 'funky' stuff in the first place as this is a tool designed for developers to generate compressed textures.

 

[Dio]

It just compares two named images.

One of the problems with any tool like this is you have to know which is 'right' and which is 'wrong'. The Compressonator neither makes this decision for you, or frankly even gives you any significant help in making it. All it will say is 'these are different' and the brain of the user then has to do the rest.

One other point: the Compressonator has a very good texture compressor in it and is an excellent tool for doing this job. It even competes well with S3's reference compressor.

 

[Graphics_Krazy]

Try using this tool to compare against different Detonator driver sets.

Also using the FRAPS movie mode is useful for lining up images. Use VirtualDub (freeware) to line up the exact frame to compare with another image.

 

[Zvekan]

download "The Compressonator".

Thnx, great proggie. Much better than constant image filpping in Acdsee

Zvekan

 

[Genghis Presley]

Woh, I was surprised to see The Compressonator make it onto Beyond3D but I was even more surprised to see conspiracy theories instantly surround it.

For the record, it has no mechanism for detecting video hardware and I can't think of any mechanism which could do this, nor what purpose this would serve. It doesn't use Direct3D or even DirectDraw.

If I had the time to do anything like vendor detection I'd be spending it instead on getting v1.1 of The Compressonator released along with it's plugin SDK. Unfortunately as TC is a side project of mine that I started of my own volition it's taking longer than I intended.

I'm glad to see that most folk like the tool, I guess you haven't noticed any of the bugs yet.

If anyone has any comments or suggested improvements I'd be glad to hear them.

GP.

 

[jb]

I'm glad to see that most folk like the tool, I guess you haven't noticed any of the bugs yet.

Let me give you some advice. When people come to me with bugs on my UT mod I tell them thats not a bug, but a undocumented feature that can have a random occurance See no bugs

Nice tool BTW.

 

[Genghis Presley]

I'm glad to see that most folk like the tool, I guess you haven't noticed any of the bugs yet.

Let me give you some advice. When people come to me with bugs on my UT mod I tell them thats not a bug, but a undocumented feature that can have a random occurance See no bugs

That's why I always try to avoid documenting anything, so everything is an undocumented feature.

Nice tool BTW.

Cheers.

GP.

 

[Sxotty]

Like they said it still means little, all it shows is difference. Reviewers still have to subjectively decide which is better, and unless they have a reference rasterizer then it is still a matter of opinion. All I am saying is that I hope other sites are smart enough to realize this so they don't say things like by comparing these two images we can clearly see that X does it wrong...