Render Quality, Is "Best" really best?

[Andrew Schreiber]

A common thought we all have is that we assume that something that is "Best" is always better than what is marked "Good". If I buy a new processor that is marked 2.6gHz that is better than my current 2.4gHz one, the first question I ask myself is, "What benefit do I gain from this?" Now we all want our renders to look as good as possible but at what point does it become counter-productive to use higher and higher settings. In an attempt to see what the difference would be for different settings I did a pair of test renders using Photoworks 360. My test machine:

Intel Core 2 Quad Q6600 @ 2.40 GHZ
8.00 GB Ram
Windows 7 64-bit
ATI FirePro 3750 Graphics Card

Results:

BEST:

Render Time: 1 hr 5 min 29 s
Size: 1920 x 1080
Threads: 4
AA Samples: 32
Indirect Rays: 2048

GOOD:

Render Time: 2 min 26.9 s
Size: 1920 x 1080
Threads: 4
AA Samples: 8
Indirect Rays: 128

As you can see there are a few distinct differences. The most distinct difference is the brightness of the resulting image, the BEST rendering is significantly brighter in places than its counterpart. This can be explained by the increase in Indirect Rays, this means that there are more rays. This manifests itself very visibly in areas where light would bounce around a lot such as the teeth of a brightly polished gear. (Left is BEST)


The other primary difference is the number of Anti-Aliasing passes, 4 compared to 32. This should mean lines that are not straight should look better in the BEST rendering. (Again, Left is BEST)


Now, the real question, can you tell which is which? link

My conclusion based on this evidence that 32.5 times the rendering time does not equal a 32.5 times better image. I will continue using the "Good" rendering setting. Obviously this is a personal choice and your mileage may vary.

The solution is that the rendering on the RIGHT is BEST. Now, as a disclaimer, all images were taken from Photoworks 360 as .jpg images. The clippings were made as similar as is humanly possible.

All images are available in full resolution by just clicking on the image and choosing Original.

[Akash Rastogi]

heh, you just wanted to show off that drive train didn't you?

Haha, nice quality check on those renders though. I've never gone to "Best" myself and its cool to see the benefits.

Neat design by the way, although I don't very much like that integrated gearbox.

[NickE]

Originally Posted by Andrew Schreiber

\The most distinct difference is the brightness of the resulting image, the BEST rendering is significantly brighter in places than its counterpart. This can be explained by the increase in Indirect Rays, this means that there are more rays. This manifests itself very visibly in areas where light would bounce around a lot such as the teeth of a brightly polished gear. (Left is BEST)

It manifests itself even more obviously in the reflections of the chassis on the surface. In the GOOD rendering, all of the bolts are black. In the BEST rendering, all are silver.

Thanks for this information. I have been working on some renders lately and was wondering exactly the question of if BEST quality was worth the wait.

[Andrew Schreiber]

Originally Posted by NickE

It manifests itself even more obviously in the reflections of the chassis on the surface. In the GOOD rendering, all of the bolts are black. In the BEST rendering, all are silver.

Thanks for this information. I have been working on some renders lately and was wondering exactly the question of if BEST quality was worth the wait.

Nick, good point there. One thing that might be worth playing with would be the brightness settings. Not sure if they will change anything but hey, rendering at BEST means I have 30 shots to get the render right eh?

Did you come to a conclusion on which quality was correct? What were your criteria?

[Molten]

I hate long renders. I would just stick to the good setting.