Downsampling with AMD: Guide and Demonstration (Antialiasing Substitute)

I'm not suggesting that they're making things up - just that the shots in the article don't do a great job of a fair comparison. I can imagine this technique does reduce jaggies and that's great...I'm just not sold on it's advantages over using SSAA if you've got the graphics card grunt to do so.

The ability to play in higher resolutions than what your monitor allows? I've tried forcing SSAA in a few games as I have said, it either kills my FPS completely or makes the game do ridiculous things with the textures and meshes.

The FPS hit of this is tiny in comparison, and gives the bonus of actually being in a higher resolution. Maybe the OP shouldn't of suggested it as an AA alternate, simply just offered a way for people to play at higher res.
 
Omaeka I'd expect a ban if a mod spots your personal rant at FrenchTart.

Like stated earlier it is probably more noticeable to you because of your low res.
 
for some reason this is only allowing me to add the new res to the right monitor, not my middle (primary) one... any ideas?

apply it for each monitor by making the one you are using the primary and disabling the second; then apply the new settings from the single enabled display.

This is what I had to do to get it to wok for my HDTV while my LCD was connected. There might be another way to do this by altering the device ID number at the top of the utility, I did not however try this.


I will add this to the FAQ.
 
The ability to play in higher resolutions than what your monitor allows? I've tried forcing SSAA in a few games as I have said, it either kills my FPS completely or makes the game do ridiculous things with the textures and meshes.

The FPS hit of this is tiny in comparison, and gives the bonus of actually being in a higher resolution. Maybe the OP shouldn't of suggested it as an AA alternate, simply just offered a way for people to play at higher res.

Not sure if you missed my post on the previous page but your problems with SSAA just sound like bad luck. Normally it's fine. Personally I don't use it as I dread to think what it would do to my fps at 2560x1440 though.

Have you compared the performance hit of this vs 4xMSAA or MLAA out of interest?
 
apply it for each monitor by making the one you are using the primary and disabling the second; then apply the new settings from the single enabled display.

This is what I had to do to get it to wok for my HDTV while my LCD was connected. There might be another way to do this by altering the device ID number at the top of the utility, I did not however try this.


I will add this to the FAQ.

nice one, thanks
 
Fair enough.

It's an alternate method of achieving higher image quality and without relying on post process methods like FXAA/MLAA. Though this method can also require lots of GPU power (and vram), it can also provide comparable image quality to SSAA.

Surely that can't be a bad thing, especially when SSAA isn't available or may have compatibility issues.

Not a bad thing at all and clearly some people like it. I probably won't use it for the same reasons I won't use MLAA but if it works for some then it's a great resource and once this thread is cleaned up I'm sure it'll be helpful to many.
 
What about GTAIV, I'm not sure if any other games do it but I know that one scales everything based on whatever resolution it is in; shadows, textures, everything.

This is a neat little trick from a 100kb download, its definitely worth a topic.
 
What about GTAIV, I'm not sure if any other games do it but I know that one scales everything based on whatever resolution it is in; shadows, textures, everything.

This is a neat little trick from a 100kb download, its definitely worth a topic.

It improves the picture quality significantly, but the performance hit isn't exactly small either.

In any case, you'd be better off using Super-Sampling than downscaling from 2560x1440 to 1920x1080 (for example).

GTA IV allows you to use this method with ENB configs, by the way.

You have to bear in mind that downsampling won't be so effective it isn't scaled in perfect ratios. You'd have to render the whole image at 4x the native resolution to get clear and sharp picture since anything less than that will result in the loss of data using some rather complicated algorithms.
 
It doesn't actually let you do that.
Look at the two screenshots.

The posted screenshots are at different resolutions. The linked site simply allows you to view both at the same viewable resolution. Right click on each image and check image information for actual resolution or save to desktop.

Not a bad thing at all and clearly some people like it. I probably won't use it for the same reasons I won't use MLAA but if it works for some then it's a great resource and once this thread is cleaned up I'm sure it'll be helpful to many.

This method is clearly not comparable to FXAA in any shape or form. The only method it is vaguely comparable to is supersampling AA. In any case, if you prefer to use FXAA/NO AA for games games where SSAA is not available then that's your choice.

To each their own.
 
This method is clearly not comparable to FXAA in any shape or form. The only method it is vaguely comparable to is supersampling AA.

In any case, if you prefer to use FXAA/NO AA for games games where SSAA is not available then that's your choice.

To each their own.

I would say that it's comparable to MLAA. Not in terms of methodology but in terms of end result - slight blurring and reduction in jaggies.
 
It improves the picture quality significantly, but the performance hit isn't exactly small either.

In any case, you'd be better off using Super-Sampling than downscaling from 2560x1440 to 1920x1080 (for example).

GTA IV allows you to use this method with ENB configs, by the way.

You have to bear in mind that downsampling won't be so effective it isn't scaled in perfect ratios. You'd have to render the whole image at 4x the native resolution to get clear and sharp picture since anything less than that will result in the loss of data using some rather complicated algorithms.

I got the exact same performance from IV @ 720p, 1440x900, 1680x1050 and 1080x1200. I don't think the game uses GPU for much more than textures and res, hence why its stupidly CPU dependent and runs like ass on anything below an i5. Even then it can be extremely random on what computers it actually works well.
 
The posted screenshots are at different resolutions. The linked site simply allows you to view both at the same viewable resolution. Right click on each image and check image information for actual resolution or save to desktop.

The exact same information is in both screens.
There's no more being rendered.
When you play in Eyefinity or whatnot, you get that extra peripheral vision, even with the "extra resolution" in those screenshots, there's nothing extra.

Sure, I won't argue its ability to remove jaggies improve IQ, but it's certainly from my point of view not adding any extra details/fov that would be associated with a higher resolution.
 
I would say that it's comparable to MLAA. Not in terms of methodology but in terms of end result - slight blurring and reduction in jaggies.

There is more than enough evidence scattered around the internet to show this is completely false.

But like I said, to each their own.


The exact same information is in both screens.
There's no more being rendered.
When you play in Eyefinity or whatnot, you get that extra peripheral vision, even with the "extra resolution" in those screenshots, there's nothing extra.

Sure, I won't argue its ability to remove jaggies improve IQ, but it's certainly from my point of view not adding any extra details/fov that would be associated with a higher resolution.

That's why it's called downsampling. Also high resolutions with the same FOV aren't meant to add more detail.
 
I got the exact same performance from IV @ 720p, 1440x900, 1680x1050 and 1080x1200. I don't think the game uses GPU for much more than textures and res, hence why its stupidly CPU dependent and runs like ass on anything below an i5. Even then it can be extremely random on what computers it actually works well.

So between 1366x768 and 1920x1200 or whatever you noticed no performance hit at all from using this?
 
There more than enough evidence scattered around the internet to show this is completely false.

But like I said, to each their own.

I'm sorry, which part are you saying is false? The comparison articles/screenshots on here are showing blur, along with the personal experiences of Omaeka showing blur on desktop.

I'm not saying that this is an awful technique or should never be used....what I'm saying is that there are tradeoffs.
 
Back
Top Bottom