Image Quality Comparison Tool

[^eMpTy^]

So I made a javascript toy to make it easier to compare image quality interactively from a web page...right now it just uses a static set of images...working on letting it compare any image you send it...thoughts?

http://www.gpureview.com/iq.html

 

[Dave B(TotalVR)]

1 word for that.

Cool

is there any way to subtract one image from the other to spot mipmap boundaries etc. easily?


Also, its a bit slow to load, might be coz ive got 3 torrents running though

 

[Tim Murray]

Works fine, loads quite zippily.

Needs more PNG and less JPEG, though.

 

[Powderkeg]

Needs more PNG and less JPEG, though.

Yes, if you are comparing Jpegs you are comparing compression algorithms, not the actual image quality of the original source material.

 

[^eMpTy^]

Well the idea is to make it so people can just pick any two images from the web and get a comparison.

The subtraction thing I really want to do, but I'm guessing I'll have to do that via php, and I don't know if my server has the appropriate GD library installed...

Other than the difference of the two images...can anyone think of a way to further improve this?

 

[DevilsRejection]

nice little site you got, are you going to do a review of the X1800? the more non-bias independents we have the better.

 

[^eMpTy^]

nice little site you got, are you going to do a review of the X1800? the more non-bias independents we have the better.

haha...I wish...the site is brand new, and I'm definitely not at that level yet...but all in good time...

 

[RickCain]

are you going to do a review of the X1800? the more non-bias independents we have the better.

LOL... you obviously have not read any of his posts on [H]'s forum in the video card sub-forum.

The site is nice though.

 

[fallguy]

haha...I wish...the site is brand new, and I'm definitely not at that level yet...but all in good time...

Nor non-bias...

 

[^eMpTy^]

Nor non-bias...

Right, bias = unprofessional. I consider the website a professional effort and therefore I leave my bias at the door. I have my fun ranting on forums like anyone else, but I feel that if I'm ever going to have a successful website, impartiality is an absolute necessity.

Oh and thanks for bringing it up. .!..

 

[fallguy]

I agree. However, not many are going to take you seriously, if you have a serious bias on forums. Good luck however.

 

[epicstruggle]

I agree. However, not many are going to take you seriously, if you have a serious bias on forums. Good luck however.

His work should do the speaking for him. I for one will judge that and not his commentary on forums.

epic

 

[Apple740]

I agree. However, not many are going to take you seriously

Why not, he and Trancendenz would be an awesome couple.

 

[^eMpTy^]

I agree. However, not many are going to take you seriously, if you have a serious bias on forums. Good luck however.

haha...man...I know that...I call it like I see it...always have...bias is all in the eye of the beholder...

 

[RickCain]

I agree. However, not many are going to take you seriously, if you have a serious bias on forums. Good luck however.

Would be fun to sit back and watch emept make the same comments on this forum as he does on [H].

 

[Reverend]

This thread (which is in the wrong forum IMO) reminds me of a time when I spent a considerable amount of time investigating and experimenting with ways to hopefully better represent what a person sees and how he can present it on the Internet, with the inherent problem of varying desktop resolutions among Internet readers.

Sorry if this is OT but I had once asked how a browser scales images : when I present a, say, 800x600 image *fullscreen* (the magic of JavaScript... no scrollbars, no nothing, an absolute fullscreen image but still via/in a browser) on a user's, say, 1600x1200 desktop in a browser window, how does that browser "enlarge" the 800x600 image to completely fill up that user's monitor (which may be a 15 or 21 incher) at his desktop rez?

An example would be to study the effects of a particular AA algo using low resolutions, grab a screenshot and present that AA'ed low resolution screenshot in a browser window *fullscreen*, regardless of an Internet user's desktop rez or monitor size. Just like viewing a game in 800x600 to that user.

Sorry if this is incomprehensible or you don't get what I mean... I may just cook up a JS to illustrate this.

 

[^eMpTy^]

This thread (which is in the wrong forum IMO) reminds me of a time when I spent a considerable amount of time investigating and experimenting with ways to hopefully better represent what a person sees and how he can present it on the Internet, with the inherent problem of varying desktop resolutions among Internet readers.

Sorry if this is OT but I had once asked how a browser scales images : when I present a, say, 800x600 image *fullscreen* (the magic of JavaScript... no scrollbars, no nothing, an absolute fullscreen image but still via/in a browser) on a user's, say, 1600x1200 desktop in a browser window, how does that browser "enlarge" the 800x600 image to completely fill up that user's monitor (which may be a 15 or 21 incher) at his desktop rez?

An example would be to study the effects of a particular AA algo using low resolutions, grab a screenshot and present that AA'ed low resolution screenshot in a browser window *fullscreen*, regardless of an Internet user's desktop rez or monitor size. Just like viewing a game in 800x600 to that user.

Sorry if this is incomprehensible or you don't get what I mean... I may just cook up a JS to illustrate this.

An interesting idea...but several things would screw it up. If they use Opera, the image would be resampled, thus destroying the aliasing. If they use an LCD, it resamples too, so that's again not what it would look like to them.

Also, going full screen in a browser has a couple limitations, like the titlebar and the start bar. You can get something almost full screen, but not quite. On top of all that, I just find javascript windows that exceed the limits of my screen in an effort to be "fullscreen" generally obnoxious...

In the end it'd be much easier to just have them download the image and view it an image viewer with a full screen mode...like the windows image viewer...

 

[Reverend]

Also, going full screen in a browser has a couple limitations, like the titlebar and the start bar. You can get something almost full screen, but not quite.

You don't understand (because perhaps you haven't seen it before) -- I meant completely *fullscreen* -- no titlebar or start bar. So, it *would be* indeed fullscreen. I remember testing this with Kristof privately a long time ago and it kinda scared him (he didn't know how to close the browser "window").

 

[Tim Murray]

You don't understand (because perhaps you haven't seen it before) -- I meant completely *fullscreen* -- no titlebar or start bar. So, it *would be* indeed fullscreen. I remember testing this with Kristof privately a long time ago and it kinda scared him (he didn't know how to close the browser "window").

Since it's pretty obvious that all browsers are going to resample enlarged images, why not just make a larger image?

Let's see...

Well, a 1024x768 PNG was 951KB. A 10240x7680 resized using Photoshop and nearest neighbor resampling is 1865KB.

Interesting.

 

[Hyp-X]

I remember testing this with Kristof privately a long time ago and it kinda scared him (he didn't know how to close the browser "window").

Alt+F4 didn't work?