[rivoli]

but appart from noise redusction i don't see anything special.

um, those cropped images i posted on page 1 don't look even close to me, look at the glasses, the 1.0 ones look like glasses, the beta ones just suck (and noise have nothing to do with it).

[hdesbois]

Jagged edges, I'm afraid, will always be here at low rez + high contrast. AFAIK, no renderer nor any post filter can avoid that without ruining the image with heavy blur. As for the blur through the glass, well, I don't know. Impossible to tell at this rez. But the big news it that light actually pass through the dielectric. But maybe there is still a problem, though. In current beta, the reflection of the shadow of a dielectric is a solid shadow instead of a light one (or no shadow at all for clear glass). That might be an effect of the point of view, but in the mirror, you can see only the light cast on the wall by the glassless (can you say that?) window.
HD

[max3d]

As far as I can see the well lit parts denoise much faster, while the dark areas don't improve that much. From a technical point of view this is not the best (as it are the dark areas which are really troublesome in the long run to fix), but in production this is great news. People see noise much faster in well lit parts of the scenes.

(and a third party denoiser can work miracles in dark areas without corrupting the image)

[Maximus3D]

Also to be noted is what it says along the bottom of each rendering, Pre-Release. Meaning not finished in plain english so there's probably still issues they're working on solving as we speak. And the rendertimes are pretty short for such a scene on such a computer. It satisfies my eyes enough

/ Max

[Micha]

Nice to see the first Maxwell 1.0 test images.

In this images here I wonder me only, why is at the outside walls more noise in the Maxwell 1.0 image as in the previous version?

[max3d]

Hi Max, I don't think the prerelease will differ from the final in these images. Late bug checking, but no feature or technical improvement to be expected this late in the game.

I noticed that the lighter parts clean up much faster, but possibly at the expense of the darker parts. See and look at the light passing through the third opening in the wall.

Could also be the result of a different lighting setup as the fourth opening doesn's show as much light as in the old version and the brightness of the originating light seems to be different. To my eyes that is and they are tired of a long day looking at a bad monitor. Go out for some beers and look at it again tomorrow.

[aitraaz]

Could one say that, essentially, the difference between the beta and the pre-release (in this case), would appear to be essentially an improvement in dialectrics? and that alone? The noise in the beta seems to be generated by dialectrics alone, and in the pre release it would appear to more or less the same, the difference being less dialectrics noise? Possible? The additional noise on the exterior of the space could be noise from more sampling from light passing through dialectrics (like 5000 pixels with pre-release as compared to 500 from beta)? - light is passing through the glass, nice

Cool 'classified image,' btw...

[Thomas An.]

I think overall this looks like a serious improvement.

In the beta image the caustic noise seems to be at an early stage. The beta image would likely need another 4-6 hours to clear (From my experiece those caustic dots are nasty).

On the other hand the pre-release image is at the end of its noise refinement. The noise in the darker areas is very acceptable and may only need another 20min to clear.

[neil hayes]

"These are simple tests to show the improvements in noise removal with Maxwell 1.0"

...very encouraging signs I would say. The reduction in noise versus the beta looks significant to me, based on these previews.

-neil

[aitraaz]

Yep its definitely a step forward...its hard to tell as concerns other areas (the room is almost entirely lit directly), but it's looking interesting...i guess if we look at it proportionally, if the small image is 40 mins, the large one would be 3-5 hrs...

[tom]

Hi Maxwellians,

It's great you mostly noticed and accepted the improvement in noise. Rendertime for the big image (800 x 600) is 8 hours @ Pentium-4 2.8 Ghz
For a single CPU at this speed, I don't think it's long. In other hand, it would surely take beta many many hours without noise and light through dielectrics would never be comparable to new engine anyway. The team is still smoothing the engine so don't worry for the minor things you've noticed. We'll post further tests about noise and speed as soon as possible. So, keep your eyes on Announcements section. You'll have outputs of new features in the following days. We'll try to provide higher resolution comparisons. This is a matter of time you know... Thanks for your critics!

Best regards,
Tom

[Mihai]

Post some new material features.....PLEEEEEEEEEEEEEEEEEEEEEAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAASEEEEEEEEEEEEEEEEE

[rivoli]

You'll have outputs of new features in the following days. Thanks for your critics!

hey tom, speaking of new stuff, what about a couple of materials pics? i'd really like to see what the new multilayers shader system is capable of.

edit:
what mihai said, pleaze.

[DELETED]

DELETED

[Pietro Spampatti]

new update on the post

P.