PDA

View Full Version : Jewelry Image Rendering Challenge.


ThomasAn.
03-14-2006, 05:38 AM
Hi all,

sorry for the long absence.

A couple of things:


This challenge is ongoing indefinately. You can still submit images now or a year from now.
Every now and then the images will be grouped and will be posted for voting. This way we can look at the past voting rounds and see what we can do to improve and outdo the earlier tries.
Round01 of voting can be seen here (http://mywebpages.comcast.net/doehtal-rx8/RenderingChallenge01/ChallengeIntro.htm) (the link at the bottom of the introduction page).
Voting rounds are also open indefinately, you can still vote any time you want for present or past images
Please keep sumbitting images (please read the guidelines too)
Try to use the Nurbs files, because I see the OBJ and 3ds mesh files are coming up with some smoothing issues
Can you outdo the images from voting round 01 ?


Challenge Home page:

http://mywebpages.comcast.net/doehtal-rx8/RenderingChallenge01/ChallengeIntro.htm

The text bellow from the original post is OUTDATED. Please use the challenge website above for instructions.
UPDATE1:

Voting will commence soon for image entries submitted between 03.14.2006 and 05.20.2006. Any new images posted will not be part of this voting round.
Hi all,

For anyone interested:
I would like to invite you on a challenge of a jewelry-type scene.

The goal of this challenge is to use ANY render engine you want and produce a high-end photoreal image. (Ideally the image should pass as a photo). However, it should also be done in the least amount of render time as possible and preferably on a single CPU.

Priority No1 is realism. Once realism has been achieved then we will look at the render times and see how long (or short) it took.

We will assume that the objects are new and have no weathering (no scratches, etc) and thus no texturing is needed for this scene. A few areas I am interested to see is the transfer of glow between the gold ring and the silver coin (on the right hand side), the definition of caustics, the smoothness of the blurred reflections, antialiasing of the burned out areas, and the overall realsim of the light distribution/ilumination.

This challenge will also serve as a comparison for speed/quality between various render engines as well as a chance to improve skill.

Again, I would like to point out that the images should be in "Final Quality" and artifact free. In other words ... something worthy for a client to put in product catalogs.

Looking forward to see your results here.

-Thomas

_________________________________
Level 1 Guidelines
http://img91.imageshack.us/img91/7972/instructions0hb.th.jpg (http://img91.imageshack.us/my.php?image=instructions0hb.jpg)

FILE (obj - 4.1mb ) :
Mirror 1: http://rapidshare.de/files/15512986/Glb-Challenge_OBJ_.zip.html (http://rapidshare.de/files/15512986/Glb-Challenge_OBJ_.zip.html)
Mirror 2: http://www.megaupload.com/?d=SLCQWLIL

FILE (3ds - 1.5mb ) :
Mirror 1: http://rapidshare.de/files/15513121/Glb-Challenge_3DS_.zip.html (http://rapidshare.de/files/15513121/Glb-Challenge_3DS_.zip.html)
Mirror 2: http://www.megaupload.com/?d=EA5TAHAE
(Note: The mesh for the rings appears to be split for some reason. You can join them back together. It was an error of the exporter. Sorry about that.)

FILE (vrml - 3.5mb ) :
Mirror 1: http://rapidshare.de/files/15513375/Glb-Challenge_VRML_.zip.html (http://rapidshare.de/files/15513375/Glb-Challenge_VRML_.zip.html)
Mirror 2: http://www.megaupload.com/?d=8GZF1VG6

FILE (3dm - 0.6mb ) :
Mirror 1: http://rapidshare.de/files/15513472/Glb-Challenge_3DM_.zip.html (http://rapidshare.de/files/15513472/Glb-Challenge_3DM_.zip.html)
Mirror 2: http://www.megaupload.com/?d=R3HXOQEX

FILE (step - 0.5mb ) :
Mirror 1: http://rapidshare.de/files/15513570/Glb-Challenge_STEP_.zip.html (http://rapidshare.de/files/15513570/Glb-Challenge_STEP_.zip.html)
Mirror 2: http://www.megaupload.com/?d=EN05A1IS

Level 2 Guidelines (Strict interpretation / advanced)
http://img490.imageshack.us/img490/2162/minimumguidelines4vm.jpg (http://imageshack.us)
1 - 'V' shaped high definition caustic on silver ring http://img490.imageshack.us/img490/9678/highdefinitioncaustic8pw.jpg (http://imageshack.us)
2 - Smooth (non splotchy) glowing blurry reflection on the silver coin
3 - Smooth (non splotchy) caustic reflection on the surface of the ground plane between the two coins
4 - Smooth (non splotchy) glowing blurry reflection on the right-hand silver coin
5 - Glow on golden ring (at location 5) interacting with location 4

holle
03-14-2006, 09:43 AM
Hi Thomas.

What is a *.7z file?
Canīt open it here. Maybe an fbx or obj file would be fine ...

Edit: Ah, ZipGenius can handle it ...

Chris-TC
03-14-2006, 10:43 AM
If you want this to work, you should

a) post it in the appropriate forum -> General Techniques - Lighting and Rendering
b) use a more widely-known compression format such as RAR (I had to download and install 7-Zip)
c) provide the objects in a format like .obj - I just tried importing your .3ds into my 3d package but all I'm getting is the coin ring

ThomasAn.
03-14-2006, 07:53 PM
Hi Chris,

Thank you for the suggestions.

Uploaded now are zipped versions of (obj, 3ds, vrml, 3dm, step)

-Thomas

holle
03-15-2006, 07:10 AM
Hi,

here my first try with finalrender2 and cinema 4d.
Scene use dof, caustics, arealights for the emitter and glossy
reflections.

http://img75.imageshack.us/img75/9601/holle014po.jpg (http://imageshack.us)

ThomasAn.
03-15-2006, 05:12 PM
Holle,

Sweet :) ... thank you for giving this a try on Final render ... and 27 min is a really good time for this scene.

Did you use blurry (non glossy) reflections for the surface of the coin ? (I see they are a bit shinny) I like to see how FR handles that aspect too :)

ThomasAn.
03-15-2006, 05:19 PM
On behalf of DavidR, here is a try with "Toucan".
Toucan is a render engine in development (alpha stage and not yet available) for RhinoV4.

http://forums.splutterfish.com//files/result.jpg
(http://forums.splutterfish.com/download.php?id=3009)

ThomasAn.
03-15-2006, 05:28 PM
Hi all,

Here is an entry from "hdesbois" and done with Maxwell.

I think it has great photorealism ! Quite impressive.
Thank you hdesbois for the entry !

Ok, that's not the fastest render (lvl21/ 21 hours), but the setup has been made in no time. The longest part was to figure out the POV. I'd be curious to see if anther renderer can match the caustics' quality.
http://img53.imageshack.us/img53/7671/challenge3un.jpg (http://imageshack.us)
HD

AlexC
03-15-2006, 06:22 PM
Can you please host is somewhere else? RapidShare is not quite ........ rapid. Every time I go to download the file, it says Invalid Session.

Jelmer
03-15-2006, 06:35 PM
interesting idea ! I don't have to much experience with shaders so i will give it a try in 3ds max using mental ray, i really like the maxwell render, but 21 hours, that's sick :P

LarsSon
03-15-2006, 06:54 PM
Hi ThomasAn.

Here is another Finalrender version. I'm not happy with gaustics. Well after hdesbois
send that Maxwell render, there can't be better lookin gaustics anymore.
But rendering time is allmost 1/10 so...

I'm cooking V-rays ppt version too. I'll show that later.

Tweaking time 2 hours. Rendering 2h 32min.
Max 8 and Finalrender stage 1 sp2d.
Pentium 3.2 - 1G ram

http://img367.imageshack.us/img367/7299/cgchallenge2h32m6og.jpg (http://imageshack.us)
[/url][url="http://imageshack.us"]http://img95.imageshack.us/img95/7559/cgchallenge2h32m0ec.jpg (http://imageshack.us)


-LarsSon

ThomasAn.
03-15-2006, 07:06 PM
']Can you please host is somewhere else? RapidShare is not quite ........ rapid. Every time I go to download the file, it says Invalid Session.

No problem :) .... added a second mirror.

ThomasAn.
03-15-2006, 07:08 PM
Here is another Finalrender version. I'm not happy with gaustics. Well after hdesbois
send that Maxwell render, there can't be better lookin gaustics anymore.
But rendering time is allmost 1/10 so...

I'm cooking V-rays ppt version too. I'll show that later.

Tweaking time 2 hours. Rendering 2h 32min.
Max 8 and Finalrender stage 1 sp2d.
Pentium 3.2 - 1G ram
-LarsSon

Thanks for entry Larson !
Yes, the caustics are not there yet, but I like the look and feel of the coins.

AlexC
03-15-2006, 08:29 PM
Ok here is my WIP using Max 8 and vRay:

http://www.thecube.zeeder.com/3D/Shader/Challenge/WIP1.jpg

Render time 7 mins 9 Seconds.

Few questions:

Whats with the huge poly count on all the objects?
There were no lights/cameras in the scene, so I put my own in - is that ok?

ThomasAn.
03-15-2006, 08:37 PM
']
Few questions:
Whats with the huge poly count on all the objects?
There were no lights/cameras in the scene, so I put my own in - is that ok?

Alex,

I went a little over on polys to make sure the coins have enough detail around the edges because that is a critical area for the high definition caustics.

About lights:
In the instruction image there are two arrows. One arrow is pointing to the small rectangular surface in the front. This is one area light.
There is another arrow pointing to a large vertical surface on the side. That is the another area light.
Please use only these two as lights

There are two more planar surfaces (one on top of the scene and one behind the small light) these two are just reflector planes.

AlexC
03-15-2006, 09:02 PM
I don't just mean on the coins, I mean the whole scene - in total it's 125,000 polys - here's a screen capture.

http://img403.imageshack.us/img403/4536/wire4wr.th.jpg (http://img403.imageshack.us/my.php?image=wire4wr.jpg)

Ok, I will re-do my attempt using those lights instead, I think i'll do that tomorrow.

Thanks

ThomasAn.
03-15-2006, 09:07 PM
Hi all,

This is my try with Vray (Irradiance Map + Light Cache).
I am not very skilled with Vray .... so this mage took about 14hours to cook and the coins are still not clean enough. I had a slight difficulty making those coin caustics look sharp enough by using IR map.

http://img89.imageshack.us/img89/3811/ringscoins4qo.jpg (http://imageshack.us)

ThomasAn.
03-15-2006, 09:35 PM
']I don't just mean on the coins, I mean the whole scene - in total it's 125,000 polys - here's a screen capture.

Ok, I will re-do my attempt using those lights instead, I think i'll do that tomorrow.

Thanks

Alex,

There is a Nurbs version of the OBJ file included. There is also a STEP file (Nurbs) and (if you have Rhino) there is also the 3dm file.

If you want you can use one of those files and make the mesh as light-weight as you wish; as long as the fine details are preserved (the coin edges and emblem and the rounded grooves on the rings)

kevmo
03-16-2006, 12:26 AM
Hello:

Just took a quick test, with no DOF but I would most likely add that in post anyway.

http://www.cornellconcepts.com/temp/EIAS/EIGI.html

The caustics don't show, needs some tweaking... :)

kevmo
03-16-2006, 12:36 AM
Those are all some very nice renders by everyone.

ThomasAn.
03-16-2006, 12:42 AM
Hello:

Just took a quick test, with no DOF but I would most likely add that in post anyway.
http://www.cornellconcepts.com/temp/EIAS/EIGI.htmlhttp://www.cornellconcepts.com/temp/EIAS/Glb-Challenge_EIAS.jpg
The caustics don't show, needs some tweaking... :)

Thanks for your ElectricImage entry Kevno !
Yup, some more tweeking for the caustics would help :) ... also the coin surface could use some more roughness (blurry-ness). :)

kevmo
03-16-2006, 01:21 AM
Thanx Thomas:

I couldn't get the image to load, it said it was too large in kbs thanx for the assistance.

Yes that is with stock settings - I haven't really tried caustics so this will give me some practice.

I usually have trouble with importing .obj files, there is still some faceting so maybe I will try the .3DS model or maybe I will just use the Encage Plug-in but I wanted to keep it as pure as possible and not add any xtra polygons.

when I get some time I'll give it more time.

Coliba
03-16-2006, 01:48 AM
kevmo, did you place lights in the right positions? In your image the little area light seems to be completely missing.

ENSLAVER
03-16-2006, 04:10 AM
ThomasAn: Nice idea, i will give it a go soon as free render time =)


I noticed the "quick instructions" only pointed to a thumbnail, so here is the full size image link:
http://img97.imageshack.us/img97/6694/instructions9km.jpg

holle
03-16-2006, 07:56 AM
Holle,

Sweet :) ... thank you for giving this a try on Final render ... and 27 min is a really good time for this scene.

Did you use blurry (non glossy) reflections for the surface of the coin ? (I see they are a bit shinny) I like to see how FR handles that aspect too :)

Hi Thomas,

here a version with more amount of blurry reflection, also I raise the samples for the
arealights.

http://img95.imageshack.us/img95/3278/holle023ak.jpg
(http://img95.imageshack.us/my.php?image=holle023ak.jpg)

ThomasAn.
03-16-2006, 08:47 AM
Hi Thomas,

here a version with more amount of blurry reflection, also I raise the samples for the
arealights.
(http://img95.imageshack.us/my.php?image=holle023ak.jpg)

Looks better !
Thanks for your entry, man :)

ThomasAn.
03-16-2006, 08:49 AM
Hi all,

On behalf of Mihai ... here is an entry using MentalRay on XSI

2h 25m (on AMD X2 4800)
http://i2.tinypic.com/rj4paf.jpg

thev
03-16-2006, 01:05 PM
Here is another one with V-Ray; I didn't bother too much to optimize the settings, will try another one with lower noise level and more DOF:

http://www.spot3d.com/vray/images/stuff/rings_caust1.png

eyebee
03-16-2006, 04:06 PM
lhere is another one with maxwell after 1 hour rendering time...
should have given the glossy parts a little bit roughness, perhaps around 0.1

rendered on a dual xeon 3.6

http://img208.imageshack.us/img208/1464/challenge1re.jpg

eyebee
03-16-2006, 04:08 PM
here is my maxwell attempt:

http://img208.imageshack.us/img208/1464/challenge1re.jpg (http://imageshack.us)

rendered 1 hour on a dual xeon 3.6

thablanchh
03-16-2006, 05:13 PM
http://www.geocities.com/thablanchh/ringtest-gamma22workflow3.jpgIn Vray
2.2 Gamma linear workflow
47min on a optron 275

ThomasAn.
03-16-2006, 05:29 PM
In Vray
2.2 Gamma linear workflow
47min on a optron 275

Hi thablanchh,

Good one :)
Are you using photon mapped caustics ? or is it IR+LC ?

thablanchh
03-16-2006, 05:45 PM
It is photons caustics

Rendered with QMC + IR MAP, lightcache as second bounce gave a little less detail, especially as I am making this scene way brighter, every little detail counts.

I went a little hard on the DOF:D

Ciao

Jelmer
03-16-2006, 07:22 PM
more nice entries ! maybe people could post some of their material settings so we can learn from it, i'm still working on my 3dsmax-mental ray render, the light and DOF is easy but i'm struggling a bit with the shaders.

ThomasAn.
03-16-2006, 07:44 PM
Hi everyone,

I would like to point out that the images should be in "Final Quality" and artifact free. In other words ... something worthy for a client and put in product catalogs.

ThomasAn.
03-16-2006, 09:37 PM
Hi all,

Here is one image using Maxwell RC5 (24hours) on AMD 64 3500
http://img205.imageshack.us/img205/5159/ringsb08g18iso100sh12024h00m2w.jpg

ThomasAn.
03-17-2006, 01:01 AM
Here is another one with V-Ray; I didn't bother too much to optimize the settings, will try another one with lower noise level and more DOF:



thev,

I like the contrast in your image and the materials are very good giving an excellent metallic feel.

Yes, a little more DOF will be better ... and if you don't mind ... a little more blur/rougness on the coin surface.

(Be carefull though ... if you make it too good, I might ask you for your settings :) )

ThomasAn.
03-17-2006, 01:05 AM
Hi all,

Here is an entry from BigK:

so here are my tries on my PIV 3.4 GHz

cinema 4d 2h35min (have to change the blurry reflections though and something is a little off with the caustics)
http://www.kiframes.de/maxwell/renderChallenge_c4d.jpg

maxwell SL 18 in 19h45min (setup in really no matter of time ;-) ) DOF is a little strong (should have used fstop 5.6 instead of 4)
http://www.kiframes.de/maxwell/renderChallenge_mw.jpg

i tried in fR2 for cinema, but something didnīt work out. so if holger reads this. please tell me how you got caustics for the area lights.

see yous
michael

ThomasAn.
03-17-2006, 01:09 AM
One entry by rivoli (using Vray)

here's my vray attempt. rendered for 50 minutes on a dual old prestonia xeon 2400, 3 gigs ram (it used up to 1,1 gig during rendering). setup time, test renders and all took me about half an hour/45 minutes.

I know you're interested in vray settings Thomas, so: qmc+lc+photon mapped caustics (adaptive subd AA). I'd say that overall we're in good shape, those caustics still look noisy/blurry but I couldn't get any further with them (I ended up with a 800mb photon map, that alone sucked a lot of memory), and the glossy coins still look bad.

with this being said, for shots like this one (small scene, you can have it done in just one pass, super realistic/detailed dof, shaders and caustics needed...) maxwell would be my first choice anyway.

http://www.caribouz.com/vray/test_rings_corrected.png

ThomasAn.
03-17-2006, 01:10 AM
An entry by dutch_designer using Brazil R/S:

Here's my final Brazil result, the big area light is weaker than my small area light, also emits less caustic photons. Result is not too shabby considering my hardware limitations and the time taken.

caustic map creation: 8 mins
render: 2hour 35 minutes

hardware:
P4 Northwood 2.8 GHz HT
1 Gb ram
physical memory used: 324,3M
virtual memory used: 391.6M
image sampling min/max: 2/2
dof enabled
no secundary lighting

http://www.student.io.tudelft.nl/io9697340/result_brazil_02.jpg

some of the Brazil Console info:


RND: Opening host initiated Brazil r/s session
EVT: HOST pre render event
RND: frame render called
MXT:
MXT: 3DSMax<->Brazil Scene Transport
MXT: Đ2001,2002 SplutterFish, LLC.
MXT:
BKT: opening renderer
BKT: image width: 800, height 600
BKT: bucket size: 32 x 32
MXT: updating scene database
MXT:
MXT: Renderable geometry:
MXT: 58 objects
MXT: 126969 faces
MXT:
MXT: Node Level Materials:
MXT: 6 unique materials
MXT: 56 Brazil Based
MXT:
MXT: Light Objects:
MXT: 2 total lights
MXT: 2 Brazil rectangle areaBKT: waiting for threads to exit...
BKT: render complete.
EVT: HOST post render frame event
RND: Closing bucket renderer
RND: Copying buffer to bitmap
LIT:
LIT: Light Statistics for "Brazil Light01" at t=0
LIT: # of premature illums: 0 (0,000000 %)
LIT: max samples/illum : 0 (@-1 ,-1)
LIT:
LIT:
LIT: Light Statistics for "Brazil Light02" at t=0
LIT: # of premature illums: 0 (0,000000 %)
LIT: max samples/illum : 0 (@-1 ,-1)
LIT:
RND: Frame render end
RND: Closing Brazil r/s render session.
EVT: HOST post render event
DAC: default accelerator shutdown
DAC: freeing space partitions
EVT:
EVT: Render Statistics
EVT: 02h:43m:15.4s total time for Brazil
EVT: 00h:07m:46.4s for Brazil initialization
EVT: 00h:00m:01.6s for Ray Server initialization
EVT: 00h:07m:44.3s for Photon Server initialization
EVT: 02h:35m:29.0s for bucket rendering
EVT:
EVT: 858037282 total ray casts
EVT: 64800000 view rays
EVT: 15993079 reflected rays
EVT: 737044203 occlusion rays
EVT: 40200000 photon tracer rays
SYS: system idle
==================================================

MXT:
MXT: updating 2 Area Lights
MXT: done updating lights
EVT: HOST pre render frame event
SCN: updating database
SCN: renderable primitives found:
SCN: 58 polygonal prims
SCN: renderable geometry found:
SCN: 126969 polygon rprims
SCN: light sources found:
SCN: 0 point lights
SCN: 2 area lights
SCN: shaders found:
SCN: 2 shaders
SCN:
SYS: system ready
DAC: default accelerator init
DAC: using single nested grid for primary acceleration
UGD: building grid
UGD: render primitives: 126969
DAC: cleanup
DAC: default accelerator ready
RND: Frame Render
PHT: caustic photon map pass 1
PHT: creating 2 threads
LIT: Brazil Light01 Emitting 100000 photons...
LIT: Brazil Light02 Emitting 100000 photons...
PHT: terminating threads
PHT: generating caustic photon map
PHT: emitting 40000000 total photons from 2 lights
PHT: 165 Mb allocated for photon map
PHT: creating 2 threads
LIT: Brazil Light01 Emitting 10000000 photons...
LIT: Brazil Light02 Emitting 30000000 photons...
PHT: terminating threads
PHT: main pass stored 4067260 photons
PHT: 4230045 photons stored
PHT: balancing kd-tree
BKT: 475 buckets in queue
BKT: adaptive refinement disabled
BKT: multithreading enabled
BKT: using 2 threads
BKT: all threads normal priority
BKT: initializing lens shader
BKT: initializing bucket samplers
BKT: rendering[/code]

vvee
03-17-2006, 03:31 AM
XSI
45 min
quad amd

You know I forgot the dof, I'll do another one.

The model is busted on the second gold ring, in the inner back area, right area.

Vee

peter_tomas
03-17-2006, 03:39 PM
Maxwell render : 10h
P4 HT Prescott 2,8 | 2 GB ram | ati radeon 9600 XT

(http://www.atom.sk/zakaznici/peter_tomas/obr/Glb_sv.jpg)

peter_tomas
03-17-2006, 03:48 PM
MAXWELL RENDER: 10h
P4 HT Prescott 2,8 | 2 GB ram | ati radeon 9600 XT

ThomasAn.
03-17-2006, 03:52 PM
XSI
45 min
quad amd

You know I forgot the dof, I'll do another one.

The model is busted on the second gold ring, in the inner back area, right area.

Vee

Hi Vee,

Thanks for trying the scene on XSI :)


About the mesh, yes I know it is split at one spot (it must have been the converter from Nurbs). Can you join it back together ? Also I think that is only a problem with the 3ds mesh.

ThomasAn.
03-17-2006, 03:53 PM
Entry from LarSon with Vray PPT:

This is my final with V-Ray ppt.

Rendering time 18.5h with 2xdualcore opteron. Wow!
If comparing to my home pc P4 3.4, it would render 6 times longer.
Total rendering time should be then 111h!!!

http://img234.imageshack.us/img234/9382/cgchallengevray185h6vg.jpg (http://imageshack.us)

Gaustics are not clear yet. In reflection they are much better than Maxwell's,
but in surface they are too noisy still.

big K: i'm rendering one more version with FR and using rectangle lights.
Gaustics are rendered beutifully and realistic decay gives alot to this kind
of lighting situation.


-LarsSon

ThomasAn.
03-17-2006, 03:55 PM
An entry from BigK using FinalRender stage 2:

so finally, here is my fR2 version (had to tweak for the caustics) RT 5h 35min (could be optimised a little i think (antialiasing and GI)

http://www.kiframes.de/maxwell/renderChallenge_fR.jpg

caustics are not possible with area lights at the moment, so this one used point lights for the caustics)

michael

ThomasAn.
03-17-2006, 03:57 PM
Deleted duplicate

peter_tomas
03-17-2006, 03:58 PM
MAXWELL RENDER : 10h
P4 HT Prescott 2,8 | 2 GB ram | ati radeon 9600 XT

http://www.atom.sk/zakaznici/peter_tomas/obr/Glb_sv.jpg

vvee
03-17-2006, 04:00 PM
Image with DOF



XSI: mental ray/final gathering

Rendered At: 800 x 540 (the first image was rendered at 1600 x 1080)

Time: 15 min

Machine: Dual 270 AMD (quad)

ThomasAn.
03-17-2006, 07:47 PM
Another entry by Larson using FinalRender:

Puuuh... and one more. Then i'm totally done.

Finalrender 10h 20m with P4 3.4.

As any other renderer, there is problems with anti-aliasing in
those extreme high contrast areas. :? But i think this is almost
best what you can get wiht FR.

ThomasAn: Thanks for this challenge. I have learned alot of Finalrender,
because i had to push everything out from it. I also learned that for this
kind of realistic renderings there is only one option - Maxwell. :wink:


http://img123.imageshack.us/img123/4346/cgchallengefr10h20m2cr.jpg (http://imageshack.us)

-LarsSon

Jelmer
03-17-2006, 09:14 PM
the renders are getting better and better, i'm still struggling with the materials but i'll post something eventually :P

ThomasAn.
03-18-2006, 01:17 AM
An entry from ernchesto using Maxwell

Ok - here's a couple from my Maxwell.
Setup was about as close to nothing as one could get - basic materials and basic emitters. Maybe 5 minutes.
The computer is an Athlon 2500 with 1.5Gb Ram

First image: 11H43Min (screen grab)
http://img95.imageshack.us/img95/1482/11h43minsl185rq.jpg
Second Image: 20H54Min
http://img205.imageshack.us/img205/3740/20h54min1951sl9mg.jpg

Thanks for this Thomas. Fun learning.

Cheers,
Ernest

D-Alexander
03-18-2006, 05:02 AM
here is my render.

3DS Max 7 Rendered with Vray 1.47.18. Time 22:23 Minutes.

http://img58.imageshack.us/img58/2825/jewlery1fm.jpg

i was planing on making it alittle more gold so i may post another version.

ThomasAn.
03-18-2006, 05:32 AM
here is my render.

3DS Max 7 Rendered with Vray 1.47.18. Time 22:23 Minutes.

http://img58.imageshack.us/img58/2825/jewlery1fm.jpg

i was planing on making it alittle more gold so i may post another version.

D.Alexander,

Thanks for your effort man ... but you changed the position of the lights ... and also there are not caustics in the image, and no DOF...

D-Alexander
03-18-2006, 06:22 AM
i was really unsure where to place the lights, alittle more explanation would really help me out alot. as for the dofi was hoping it wouldn't be required but i will add it next time.

AlexC
03-18-2006, 08:25 AM
I'm with D.Alex - You don't specify really where you want the lights, you said those plains were lights in the scene but all they are is just a plain - no light. That's why I had to place my own lights in.

I think you should get some people to make a native format for each program you want and get the camera/lights/dof-focal point the same. This would make it a little more fair aswell when it comes down to Render Time.

Also if you want us to make it the most realistic image, why not let us choose where to put the lights?

LarsSon
03-18-2006, 06:04 PM
About scene setup.
Because this challenge is made to compare different renderers,
it is very important to have these basic rules. Exactly same sized area lights and reflectors in same place.
Same camera angle and fov. Every scene must have gi, gaustics, blurry reflections and dof on.
Now when we all render this scene with this kind of setup, we are really able to compare
what different kind of renderers can do or can't do.

I don't understand why this is problem. ?


ThomasAn: i hope i did get it right?


-LarsSon

kevmo
03-18-2006, 07:19 PM
Rendered with Electric Image
17 minutes 25 seconds

http://www.cornellconcepts.com/temp/EIAS/Glb-Challenge_5.jpg

thev2
03-18-2006, 08:51 PM
Two more with V-Ray, as a test for the "universal" V-Ray settings.

The first one is with path traced caustics - slow, of course, but probably not too bad for a 30 seconds load-a-preset-and-render setup without any tweaking:

http://www.spot3d.com/vray/images/stuff/univ_settings/rings_caust3.png

The second one is the same setup but with photon-mapped caustics:
http://www.spot3d.com/vray/images/stuff/univ_settings/rings_caust4.png

Next will be some optimizations to reduce render times, but I need to find the time for this...

Renzsu
03-18-2006, 09:03 PM
Here's my second attempt, this time with Maxwell RC5

SL 18.67, 23h54m
P4 2.8 Ghz HT, 1Gb ram

I let it render overnight but to be honest, after about 3 hours the image was good enough I think to denoise (or if I rendered in a higher res, denoise/resize).

http://www.student.io.tudelft.nl/io9697340/result_maxwell_01.jpg

ThomasAn.
03-18-2006, 09:15 PM
About scene setup.
Because this challenge is made to compare different renderers,
it is very important to have these basic rules. Exactly same sized area lights and reflectors in same place.
Same camera angle and fov. Every scene must have gi, gaustics, blurry reflections and dof on.
Now when we all render this scene with this kind of setup, we are really able to compare
what different kind of renderers can do or can't do.

I don't understand why this is problem. ?
ThomasAn: i hope i did get it right?
-LarsSon

That is exactly right !

I am a little surprised of this being an issue.
In the first page there is an image with basic guidlines:
http://img91.imageshack.us/img91/7972/instructions0hb.th.jpg (http://img91.imageshack.us/my.php?image=instructions0hb.jpg)

The red lines on the top view represent the locations of the area lights. If you open the scene you will see two surfaces corresponding to those locations.

Either turn those surfaces into area lights or replace these two surfaces with area lights of the exact same size and position.

Again: The coin surface should have roughness, DOF should be on and caustics should be on. There is also a dummy ribbon surface indicating the direction from camera to target (you can use that as a guide to place your camera... I think I included that in most of the files, if you don't see it in one file; please try one of the others)

ThomasAn.
03-18-2006, 09:16 PM
Larson,

I see your meshes are correctly rendered without distortions ... which file did you use ?

D-Alexander
03-18-2006, 09:27 PM
this is a problem becuase the instructions only appear as a thumbnail exept for this link

http://img97.imageshack.us/img97/6694/instructions9km.jpg

so i cannot see the other part of the instructions, is this only me? or are others having this problem?

edit, the problems is with firefox not displaying images corectly, this is fixed in i.e.

i will redo my entry.

ThomasAn.
03-18-2006, 09:46 PM
this is a problem becuase the instructions only appear as a thumbnail exept for this link

http://img97.imageshack.us/img97/6694/instructions9km.jpg

so i cannot see the other part of the instructions, is this only me? or are others having this problem?

edit, the problems is with firefox not displaying images corectly, this is fixed in i.e.

i will redo my entry.

Alex,

You can click on that thumbnail man :) (that is why it is there)

D-Alexander
03-18-2006, 10:25 PM
yea i know that but it doesn't display the thumbnail in ff it shows a link and the link only has the thumb.

is there a specific amount of dof you want or do you just want a generalized approch ?

anyway i will post my update soon.

great comp thomas! :)

ThomasAn.
03-18-2006, 11:47 PM
yea i know that but it doesn't display the thumbnail in ff it shows a link and the link only has the thumb.


I see.
But it is strange though ... when I click it here I get the full image in FF 1.501
What version of FF are you using ?


About DOF: The amount is not as strict ... so I will leave up to you as long it is visible

Thanks for taking a shot at this man ... and sorry about the issues with the thumbnail and stuff (I didn't realise it would cause a problem)

D-Alexander
03-19-2006, 01:34 AM
thanks for the info.

i'm using FF 1.0.4 so that may be my problem, i'll update it later and let you know if it's just that.

i think this is a great idea this may put an end to the which render is better arguements although probably not ;)

keep up the good work everyone !

LarsSon
03-19-2006, 07:38 AM
ThomasAn: I used 3ds version. When i started there wasn't any other.
Polycount is huge, but that's why it looks so good.


-LarsSon

ThomasAn.
03-19-2006, 08:22 AM
Entry by hdesbois using Lightwave 8.3 :

My attempt with Lightwave 8.3 :
http://img72.imageshack.us/img72/1556/testlw6qf.jpg (http://imageshack.us)
I'm sure it's possible to get better result, but setup time is long. For this kind of render, I'd prefer to use Maxwell. I'm on a laptop now, so, rendering time is not relevant. I may have another Lightwave try next week when I'm back on my main coputer.
HD

Tıax
03-19-2006, 11:14 AM
hi guys ,
nice idea and challenge :) and some very nice pics too. Here is my pic. not bad i think but i'm really bad at materials :( especially gold.

http://img236.imageshack.us/img236/3434/rendercont37ve.th.jpg (http://img236.imageshack.us/my.php?image=rendercont37ve.jpg)

a darker version , i like this more cuz of the silver-ring look cool :D

http://img229.imageshack.us/img229/3217/rendercont29xz.th.jpg (http://img229.imageshack.us/my.php?image=rendercont29xz.jpg)

prob i'll post 1-2 new render tonight , if i manage to make a good gold-mat ofc :) anyway , cya for now :)

ah all done with mental ray btw :)
render time is around 1 hour each , not sure cuz i was sleeping when it's done :P

vvee
03-19-2006, 02:47 PM
Here is a correction to my last image, which did not have shadows

Time: 45 minutes

XSI: MentalRay/Final Gathering

Dual core AMD 270

Note: If the client is going to be printing these images in the jewelery catolog at a physical size of 2.66 inches by 2 inches then the render times are valid. But If the client wants to make an 8.5 inch by 11 inch ad for his product, then the times have to be extrapolated.

My 45 minute 800 x 600 image will print at 2.66 x 2 inches at 300 PPI.

1600 x 1200, 5.33 x 4 inches 180 minute render time

3200 x 2400, 10.66 x 8 inches 720 minute render time


All this to say, you guys have to be kidding with 20 hour render time for a 2.66 x 2 inch image. it would take 320 hours (2 weeks) to get to 10.66 x 8 inches!

Renzsu
03-19-2006, 03:05 PM
All this to say, you guys have to be kidding with 20 hour render time for a 2.66 x 2 inch image. it would take 320 hours (2 weeks) to get to 10.66 x 8 inches!

Yeah, and it would take years to make one big enough to cover the Reichstag... what's your point? The challenge was making an image of 800x600 pixels. If a bigger image was necessary I'm sure people wouldn't sit on it for 2 weeks but find workarounds or switch render engine.
I mentioned with my picture that the quality would be good enough after rendering a few hours, but that I let it render longer to see how much less the noise would get.

vvee
03-19-2006, 03:27 PM
Yeah, and it would take years to make one big enough to cover the Reichstag... what's your point? The challenge was making an image of 800x600 pixels. If a bigger image was necessary I'm sure people wouldn't sit on it for 2 weeks but find workarounds or switch render engine.
I mentioned with my picture that the quality would be good enough after rendering a few hours, but that I let it render longer to see how much less the noise would get.

Just pointing out the not so obvious. for anyone that doesn't work in the Adv/Print industry.

ThomasAn.
03-19-2006, 04:45 PM
Hi everyone,

Up to now (first phase) the challenge did not have rigid quality requirements, so as to allow some room and see how far the artist will choose to go before considering an image to be "good enough".

Now it is time to tighten the screws a little:

SECOND PHASE:

MINIMUM JUDGING GUIDELINES
http://img490.imageshack.us/img490/2162/minimumguidelines4vm.jpg
1 - 'V' shaped high definition caustic from the edge of silver ring http://img490.imageshack.us/img490/9678/highdefinitioncaustic8pw.jpg
2 - Smooth (non splotchy) glowing blurry reflection on the silver coin
3 - Smooth (non splotchy) caustic reflection on the surface of the ground plane between the two coins
4 - Smooth (non splotchy) glowing blurry reflection on the right-hand silver coin
5 - Glow on golden ring (at location 5) interacting with location 4

thev2
03-19-2006, 08:03 PM
To screw things even more, you could also add

6. smooth reflections of caustics on the back side of the silver ring and
7. bright and glowing hilight reflections on the back side of the golden ring

vvee
03-20-2006, 02:14 AM
Thomas how can you consider sharp or blurry caustics in an image and yet, not take this obvious error into consideration? This looks like a composite of two different renderers.

ThomasAn.
03-20-2006, 02:39 AM
Thomas how can you consider sharp or blurry caustics in an image and yet, not take this obvious error into consideration? This looks like a composite of two different renderers.

Hi Vee,

No the image is not a composite.

Yes, the noise at the back of the ring is considered a defect.

Tıax
03-20-2006, 05:57 AM
hi guys ,

thomas , man dont u require too much things? v shape caustics , same glow in gold etc. i mean these thing makes pic nearly %99 same.then what can we chance and play on. for example , i made table more reflective, used gamma corrc. , used weak light sources etc so i didnt get and "v shape caustics" at the table. and no chance to get that glow on gold ring if u made other light source ( the one at the left) stronger. that makes pic unrealistic? if all pics will be same %99 , why u need ppls to render? i mean who'Ll be the winner ? mentalray-vray or the person who made it?

anyway , still ok for me if u want this way , but it's way too strict just wanted to say that.

cya and gl to everyone.

LarsSon
03-20-2006, 06:48 AM
WOW!!! Is there going to be a winner?!

...i don't think so. It's challenge not contest.
And i don't think it's about who is better artist,
it's about renderers. So, this time math will win,
not the artist. :shrug:


-LarsSon

ThomasAn.
03-20-2006, 07:26 AM
Hi all,

I am cool with what you are saying... However, it would make little sense for me to srart a multi-engine chanllenge if there was no "engine to engine" aspect.

What we could do is split this in two components and make two sortings
a) Loose interpretation (for Artist-to-Artist challenge)
b) Strict intepretation (for engine-to-engine challenge)

... but.... I am most interested at (b). My fear is that some might choose to go with (a) not because they mean to, but as a way to bypass engine limitations (or other problems).

I really wanted to see the various engines at full throttle on this one. Now if you start placing emitters arbitrarily, dimming them down, or placing them on the back so that you don't have to show caustics, or conceal other things... then the challenge is watered down, it is no longer a challenge (and it is supposed to be difficult)

Tıax
03-20-2006, 07:38 AM
hmm as i said no problem for me but i was just a bit suprised when i saw your last "rules" post :) prob because i thought it's an artist-artist (or just realism) challange. just a bit misunderstooding.
it still sounds good , i'll try to do as you want when i got time . good luck and cya all :)


ps: i was trying to mean "best quality/time" when i said "winner" , larsson ;)

LarsSon
03-20-2006, 08:10 AM
Tiax: You are right about the winner thing.
And about the best quality/time. I think people should post machine specs too.
Like my vray rendering was just fine before i calculated that rendering time
to P4 3.4. I know it's about 6x slower than my machine in work, but that
time will give more equal numpers to compare. Huh... so hard to explain
it with my english, but i hope you understand what i mean.

ThomasAn: Are you planning to put poll out of these? One for the quality and one
for the "usability"?


-LarsSon

ThomasAn.
03-20-2006, 08:22 AM
ThomasAn: Are you planning to put poll out of these? One for the quality and one
for the "usability"?
-LarsSon

Yup, I have been thinking about a poll (possibly two; as you mention)

Are we allowed to make polls here ? Can we make polls with more than 10 entries ?

abgrafx3d
03-20-2006, 11:05 AM
Here's my first attempt:

Maxwell RC5 - 6hr51min, SL 17.44, Dual 3.6gHz Xeon

http://www.abgrafx.com/maxwell/render_challenge01.jpg

Gonna make an attempt with Renderman for Maya next.

EDIT: Renderman for Maya DOES NOT support caustics in current release so I won't make that attempt. :sad:

kevmo
03-20-2006, 11:08 AM
ElectricImage

42mins 26secs

http://www.silvastore.com/temp/EIAS/Glb-Challenge_7.jpg

vvee
03-20-2006, 01:50 PM
Hi all,

I am cool with what you are saying... However, it would make little sense for me to srart a multi-engine chanllenge if there was no "engine to engine" aspect.

What we could do is split this in two components and make two sortings
a) Loose interpretation (for Artist-to-Artist challenge)
b) Strict intepretation (for engine-to-engine challenge)

... but.... I am most interested at (b). My fear is that some might choose to go with (a) not because they mean to, but as a way to bypass engine limitations (or other problems).

I really wanted to see the various engines at full throttle on this one. Now if you start placing emitters arbitrarily, dimming them down, or placing them on the back so that you don't have to show caustics, or conceal other things... then the challenge is watered down, it is no longer a challenge (and it is supposed to be difficult)

Don't take this personaly, but this is a bit ridiculous. If you want an asthectic challenge that's fine but if you want a scientific challenge this is the wrong place. It is impossible to have a scientific comparison of various aspects of renderers if ALL the conditions are not the same for every rendered image.

1) Light intinsities
2) Material Values
3) Camera position, rotation (for reflection positions)
3) Same Computer (for time comparison)

It would be more effective to simply make a list of various render features.

If photo-real is your true criteria the "control" must be a photograph, and not an image that is itself "photo-real" flawed.

It may be difficult to generate a reference photo of this set-up becouse computer caustics are generated from point light sources (might be wrong). But I know in real life a true point light source does not exist. Anyway I have never seen completly "smooth" caustics from metal ever in a real photo or nature (liquid, smoother, but not razor sharp).

ThomasAn.
03-20-2006, 04:00 PM
It is impossible to have a scientific comparison of various aspects of renderers if ALL the conditions are not the same for every rendered image.

Not entirely true. This particular image has only two main aspects
a) Presence and form of reflective caustics
b) Apperance of silver and gold metals

Therefore the materials are already narrowed down and the emitter intensities (although important are not as critical as their position (even at lower intensities the caustics will show; which is the most imprtant). Also, in initial the instructions of this challenge, it was already mentioned to use the same camera position (and there was a guide surface included).

About computer: A single CPU requirement is reasonable. It is not possible to organize a lot of people and expect them to have the same PC.

If photo-real is your true criteria the "control" must be a photograph, and not an image that is itself "photo-real" flawed.
...It may be difficult to generate a reference photo of this set-up becouse computer caustics are generated from point light sources (might be wrong). But I know in real life a true point light source does not exist. Anyway I have never seen completly "smooth" caustics from metal ever in a real photo or nature (liquid, smoother, but not razor sharp).

Again not true.
From a physcis standpoint, the placement and location of the caustics in this scene should be a single solution. in other words, all engines should converge to the same locations and caustic shapes (for that particular emitter set). We already have two engines agreeing to a great degree. So I beg to differ that a photo is not necessarily needed for the geometric aspect of light. For example we do not need a photo to tell us if Vray or Maxwell are doing a good job rendering a concave (or convex) mirror. We can use the mathematical models in lieu of reality and calculate the focal points and the like. http://hyperphysics.phy-astr.gsu.edu/hbase/geoopt/mirror.html#c1
(I have personally done some test on both Maxwell and Vray on simple geometric optics and the result correspond to the analytical models .... the same models that are used to design the very lenses of your actual camera that you would use to snap a photo in the first place).

I have a feeling that partly you might be hesitant to believe that a serrated coin will produce a caustic pattern like that.

However, if that is the case, keep in mind that we have all from time to time been adamant about the inaccuracy of an image only to be stumped by the way nature works.

And, yes a serrated coin will indeed produce intermittent caustic patterns:
http://www.joerch.org/coins/eu-10c-r.jpg
The above coin does not have the exact same geomtery as the coin in this test, but they are similar. (also in the above photo the size of the emitter is not as small as in the test ... the smaller emiiter will produce higher definition caustic)
http://i3.ebayimg.com/01/i/03/30/9a/bd_10.JPG

About realism: As a first step we should see if we can at least get past some basics (such as the presence of caustic in the first place, as well as definition of caustic). The rest of the realism would most likely depend on each renderer's implementation of materials. For examlple in Maxwell the materials are based (at least partially) in spectral data, so the golds and silvers are pretty close to a physical speciment of the lab that produced the data. Other engines allow the users to define their own materials (and this inctroduces the human error in the picture). Overal, in this particular challenge the materials are simple metals (known metals) so it might not be to difficult for the users to come close.

In the photo bellow the placements of the light source is close but not indentical to the model (this challenge is not meant as a scientific experiment), but the elements of the caustic are discernible. The outer surface of ring in the photo is heavily scratched and the reflected caustic mirrors all those zillion microscratches. On the other hand the inner side of the ring has a more smooth finish and the caustic is more smooth ... as logic dictates, had the outer surface been smooth (mirror finish as in the model) the caustic would also be ultra smooth as in the Vray and Maxwell images)
http://img128.imageshack.us/img128/9099/photocaustic6jt.th.jpg (http://img128.imageshack.us/my.php?image=photocaustic6jt.jpg)

The basic features we can deduce from this photo are that the ring should have a primary and a secondary inner caustic from the small emitter. It should also have reflected light around the outer surface and this outer caustic should be smooth in our case (because we are not using textures and the modelled ring is not scratched at all) and there should be a dark band between the ring and the reflected caustic.

Although, I am fairly new to Vray, but as far as Maxwell is concerned, its correctness in terms of refraction indices and caustic callibration has been confirmed (mostly for the beta1.2.2a engine) in a number of independent and more precise photo-vs-render experiements.

ThomasAn.
03-20-2006, 04:17 PM
Here's my first attempt:

Maxwell RC5 - 6hr51min, SL 17.44, Dual 3.6gHz Xeon

Gonna make an attempt with Renderman for Maya next.

Hello Wayne :)

That RC5 looks excellent :thumbsup:
Thanks for giving it a try .... and yes a renderman entry would be nice to see !

ThomasAn.
03-20-2006, 04:18 PM
ElectricImage

42mins 26secs



Hi Kevno,

Thank you for trying this with Electric Image :)

kevmo
03-20-2006, 05:37 PM
Thomas:

no problem, this is a interesting simple experiment.

I'd like to see some render times also without DOF...

In my case the image with caustics takes around 5min to render.

With DOF, even at a very subtle settings is where the big hit is.

And I imagine each program takes a different approach and time to render regardless of materials or caustics.

thanx

ThomasAn.
03-20-2006, 08:58 PM
New entry from glebe digital using Lightwave 8.5

I'm so surprised how poorly Lightwave has performed in these tests, I was certainly expecting better.

http://www.glebedigital.co.uk/finalLW.jpg
Ok maybe it's not terrible, but it's not going to 'make the grade' on these test conditions and those Mental/Vray images just leave it for dust.........set-up time is hideous [full day of fixing/tweaking] and the caustics are not convincing in the least.

http://www.glebedigital.co.uk/sl21.jpg
This is the final Maxwell treatment but is still cooking as it needs to clean up on the inside faces of the two rings........hopefully by tomorrow morning it'll be complete. :? :roll: :wink:
Thanks for the test Thomas, it's been interesting. Will this last M~R image pass the criteria ok? :)

ThomasAn.
03-20-2006, 08:59 PM
New entry from hdesbois using Lightwave

My new attempt with Lightwave...
http://img484.imageshack.us/img484/7836/lwnew1ye.jpg (http://imageshack.us)
Some nasty atefacts due to digital confusion post DOF (still on a laptop). I'll post a new image a soon as I'm back on my main computer. Blurry reflections made with procedural bump + G2 shader. Still not quite happy with the gold material. I'd like to have the advice of the Lightwave masters of this forum on this case (now entirely with gradients).
HD

ThomasAn.
03-20-2006, 09:01 PM
Hi all,

I suggest you try using the 3ds files.

It seems the OBJs (although exported with identical settings as the 3ds) seem to render with distortions and this will affect caustics.

MrPositive
03-22-2006, 05:18 AM
Hi Thomas

Do you want Global Illumination and Final Gather both or just GI from Mental Ray? I also don't believe that Mental Ray has Area Lights that cast photons (though I believe you are talking about spotlights that can become area lights).....I could be wrong I guess. As well, why not HDRI, which is claimed as the best method for realism? I mean you could blur the reflections enough so you cannot tell what's in them, and get more variety of reflections.
By the way I think this thread is brilliant...so there.

ThomasAn.
03-22-2006, 06:53 AM
Hi Thomas

Do you want Global Illumination and Final Gather both or just GI from Mental Ray? I also don't believe that Mental Ray has Area Lights that cast photons.....I could be wrong I guess. As well, why not HDRI, which is claimed as the best method for realism? I mean you could blur the reflections enough so you cannot tell what's in them, and get more variety of reflections.
By the way I think this thread is brilliant...so there.

Well, you can use whatever mechanism you feel will produce a satisfactory image (Something that you could sell to a client). Besides, I know nothing about Mental Ray, so I wouldn't know what advice to offer you in that aspect.

About HDRI: No, please, lets keep the relfections simple; as they appear from the scene itself.

-Thomas

thev
03-22-2006, 06:26 PM
This is probably as good as it gets with V-Ray as far as caustics are concerned:

http://www.spot3d.com/vray/images/stuff/rings_caust_final.png

There is still some noise in the reflection of the gold ring on the coin... trying to correct this now with higher settings.

ThomasAn.
03-23-2006, 06:51 AM
This is probably as good as it gets with V-Ray as far as caustics are concerned:

There is still some noise in the reflection of the gold ring on the coin... trying to correct this now with higher settings.

woo http://www.diegotorres.com.ar/mensajeitor/foro/caritas/erstaunt090.gif this is one awesome Vray render (IMHO) !
It definately beats an earlier try of mine in both quality and time ... (5h is pretty darn good)

What are the specs of your machine ?

slobodan_h
03-23-2006, 03:21 PM
thev,

Very nice render. Finaly somebody renders it with something different then Maxwell. :twisted:
I'm trying to get mental ray version out but have no time at the moment.

Nice test ThomasAn.

Slobodan

ThomasAn.
03-24-2006, 05:46 AM
Another try from hdesbois using Lightawave,

This is my last attemps with Lightwave :

http://img129.imageshack.us/img129/9845/classicehigh4fu.jpg (http://imageshack.us)

Rendertime is four hours, with 17 pass antialiasing, which is probably an overkill. Still room for improvement, but it would require better skills than mine. Area emitters, 0.1% caustics, no image processing, classic enhanced high AA with classic reconstruction filter.
HD

ThomasAn.
03-28-2006, 02:55 PM
thev,

You mention using the new Vray universal settings, but when I try them the result is not as good. What settings did you use for the Light Cache ? 1000 ? 5000 ? Did you also "Use Light Cache for glossy rays" ?

-Thomas

threedeworks
04-03-2006, 12:56 PM
here my try with LW 8.5 and fprime 2.1.

fprime officially does not support caustics, but i did the trick to use very high intensity specular-only lights to generate them.

edit: see this link for more about this trick: http://forums.cgsociety.org/showthread.php?t=241303

2 hours of rendering time on a mac G5 quad. there's still some grain/ antialiasing problems in the render which probably could resolve with longer render times or better light quality settings/ tuning. if i have time, i'll do more tests.

markus

http://img95.imageshack.us/img95/7373/cgtalkjeweleryfprime2h5tc.jpg

Jeet
04-10-2006, 02:05 PM
hi friend...

I want to know.... the challenge is over or still continueing ?

regards

jeet.

threedeworks
04-10-2006, 02:18 PM
hi friend...

I want to know.... the challenge is over or still continueing ?

regards

jeet.


hi! nice to see someone posting here. yep - seems we are a bit alone... hello... anyone else?

Jeet
04-10-2006, 02:19 PM
hi...

i hope the challenge is not ended yet...

i am gonna create one render too with AR 2.5

regards

jeet.

ThomasAn.
04-10-2006, 03:03 PM
Hi all,

Although it seems to be slowing down, the challenge is still on.
(It is a difficult scene, so I am allowing everyone plenty of time)

Regards,
Thomas

Saeed
04-16-2006, 12:53 AM
http://img227.imageshack.us/img227/2117/5f55a47ea182516a72cda6b6f21d87.jpg

This is my test in Brazil r/s.
3h 41m ( du to super smooth glossy - can be alot faster with a bit noise )

Single Athlon XP 1700
512 MB Ram

my pc isn't mush fast so i'm not enable dof. sorry don't have time to do it.

ThomasAn.
04-16-2006, 06:34 AM
Thanks for trying the scene Saeed :)

I really like how the coin surfaces turned out in your image (very smooth) and your gold material is quite successful !

About the caustics: Are you using point sources for lights ?

piajartist
04-16-2006, 06:36 AM
all of these renderings are quite impressive. The lighting and edge clarity is awesome!

Saeed
04-16-2006, 10:59 AM
thanks ThomasAn
yes spot-light.
it must be area ??
may i do it again of this challenge continueing.

olbo
04-17-2006, 09:19 PM
Just for fun, rendered with Indigo v0.4. ;o)

http://img130.imageshack.us/img130/1351/renderchallengeringsppindigo5.jpg

100h on AMD64 3000+@2,6Ghz / 2GB ram

Some PP done. (denoise, exposure, gamma, curve)


I'm waiting for bidirectional MLT. 8)

nb: I could have stopped the rendering at 24h but I wanted the
caustics to clear up more. :P :D ;o)

edit: thread@Indigoforum (http://www.flipcode.dxbug.com/board.php?topic=311)



take care
Oleg

daydreameroperter
04-19-2006, 03:48 AM
hey is this still going?

Jeet
04-19-2006, 03:53 AM
YEs man...

ThomasAn.
04-19-2006, 05:50 PM
hey is this still going?

Yup, it is still on.

ThomasAn.
04-19-2006, 05:54 PM
Just for fun, rendered with Indigo v0.4. ;o)

100h on AMD64 3000+@2,6Ghz / 2GB ram

Some PP done. (denoise, exposure, gamma, curve)


I'm waiting for bidirectional MLT. 8)

nb: I could have stopped the rendering at 24h but I wanted the
caustics to clear up more. :P :D ;o)

edit: thread@Indigoforum (http://www.flipcode.dxbug.com/board.php?topic=311)



take care
Oleg

Hi Olbo !

Thank you for trying this with Indigo (the more diverse the engine collection we got here the better :) )

100h is quite a bit but the image is not too shabby :thumbsup: (quite good actually)

daydreameroperter
04-20-2006, 01:21 AM
I'm doing pretty good on it. But certain things that should make a difference don't. like when I click emit photons from my lights on or off I get the same render. When I click final gather, and global illumination on or off I get the same render. And I do have use ray trace shadows on all my lights turned on, I'm rendering with mental ray, I have ray trace on in the render globals. Does anybody know why this is happening?

reidar
04-21-2006, 02:02 PM
Here is another from Electric Image:

http://privat.bluezone.no/reidar.richardsen/GICausticsColor.jpg

It renders in 1hr15 on one of my 2GHz G5 CPUs. I did the DOF in Photoshop based on the z-buffer. EI's built in DOF is multiframe-based, which is too timeconsuming for GI work!

Regards,
Reidar

buggsy
04-23-2006, 10:12 AM
Excellent job.

I am an EI user as well and have been trying my hand at this challenge.

Are you using the lightplane shader for this image of are they radial or parallel lights? Are you using any 3rd party shaders?

I think your results are fantastic! I am getting similar results but seem to be missing the reflection quality you are getting and the clarity of the "v" shaped caustics. Have you put the model through any specific modeler to correct the model to add in your results?

Many Thanks
Buggsy

reidar
04-23-2006, 07:46 PM
Thank you :-)

No 3rd party shaders involved. All materials are quite simple. Reflection blur for the coins, caustics on a value of 0.4. You don't want them two blow out. To get the fine definition, you must have enough photon density.

The geometry is run through Transporter, with all options ticked off to keep the original normals etc.

I do use lightplanes. 49 rays from the small plane, and 100 from the large. There is a spotlight at the same position as the small plane, with illumination and highlights turned off, casting the photons for caustics at a density of 400. And two more spotlights in the same positions as the lightplanes for GI secondary lighting.

What puzzles me is the soft caustics from the other light. They actually come from the lightplane itself! Even if I turn off caustics in the render window. This is not documented anywhere, and I'm still investigating...

Regards,
Reidar

slipknot66
04-24-2006, 06:15 PM
Here is my test, Maya/Mentalray.

slipknot66
04-25-2006, 11:35 AM
Here 2 new versions, rendered with Maya/Mentalray, with an Athlon XP 1.6ghz 512mbram
800x600 - 44minutos 32 seconds
1280x1024-4hours and 12 minutos

http://img258.imageshack.us/img258/6954/caustic800x6004vt.th.jpg (http://img258.imageshack.us/my.php?image=caustic800x6004vt.jpg)

http://img264.imageshack.us/img264/905/caustic1280x10248cc.th.jpg (http://img264.imageshack.us/my.php?image=caustic1280x10248cc.jpg)

The 640x480 version i dont remember the render time.

daydreameroperter
04-25-2006, 09:02 PM
those r looking good slipknot66

Saeed
04-30-2006, 12:10 AM
This is new and the last my work on this scene ( don't have time to do more )
this rendered in Brazil r/s with geometry lights [ without photons ] and caustic generated are
indirect castics ( GI castics ).
rendered in 10h : 20m
on Athlon XP 1700+ with 512MB RAM
image rendered in OverSapmling mode.

http://usera.imagecave.com/nDman/bb89776ac717fbc6582c5767788bdc.jpg

Saeed
05-01-2006, 11:40 PM
Challenge is finished ?

ThomasAn.
05-02-2006, 12:39 AM
Nice image Saeed :)
It has an interesting low-key mood to it.

Challenge is finished ?


Nope.
Keep it coming... and if anyone feels they want to further improve on a previous attempt, please feel free to post here.

ThomasAn.
05-07-2006, 07:00 PM
Hi all,

We should start wrapping this one up. (I will give it one more week and thats it ... we go for the voting on it)

********************************************
Kevno, thev, thev2, Tiax, if possible could you provide the specs of your machines for your entries (couldn't find that info, unless I missed it). Many Thanks :)
********************************************


Also, again many thanks ! to all who took part in this test / challenge so far :)

Thomas

Jeet
05-07-2006, 07:07 PM
Dear thomas,

How are you? i hope you remember me. :)

If you dont mind.. i am submitting the last entry.. i need some enhancement... but i learn so many things doing this test render..

http://img398.imageshack.us/img398/9883/render015ri.th.jpg (http://img398.imageshack.us/my.php?image=render015ri.jpg)

Rendered with Maxwell 1.0.
Rendering time : 4h 22m
PC configuration : AMD 3200+ 1GB RAM
Sampling level : 15.16

regards

Jeet.

ThomasAn.
05-07-2006, 07:55 PM
Hi Jeet,

Thanks for your entry !
One comment about your image: The IOR of the metals seems to be way too low and they look dark (you should try Nd=20 for both silver and gold parts)

Alternatively you should use the high precision Full IOR files (in the material editor ... and navigate to materialDatabase-->IOR files-->Others and pick (Ag) for silver and (Au) for gold)

Jeet
05-07-2006, 07:59 PM
Thank you ...

Yea it seems to be just 7.. oops..

regards

jeet...

Jeet
05-08-2006, 05:29 AM
Thamas,

Well, thank you for a note about ND number...

Yesterday night i was having a time to do that change and you can consider following image as my last entry... sorry...

I know some artifect is there in it but you know the time was limited .. ha ha ha... anyways... i learned so many things.. it was a nice fun and learn challenge...

Hope you to organize some of the new challenges soon....

http://img71.imageshack.us/img71/2209/render060nd.th.jpg (http://img71.imageshack.us/my.php?image=render060nd.jpg)

Rendering done with : MAaxwell 1.0
Render Time : 7hrs 55min
PC : AMD 3200+ 1GB RAM , low priority rendering

Regards

JEet

ThomasAn.
05-21-2006, 04:54 AM
Hello everyone,

This challenge has run for a couple of months. So this is a good point to close this round and start the voting.

A voting system is being prepared here but it is not finished yet (goto the bottom "Participant Entries and Voting" and follow the "Round01" link)
http://mywebpages.comcast.net/doehtal-rx8/RenderingChallenge01/ChallengeIntro.htm

The question I have now is this:

a) Would you like it set up so that each person is allowed ONLY ONE image pick out of the 50 ?
b) Would you like some more freedom and allow multiple image picks per person ? (we are still going to get a voting distribution since most people's slections will be different than others)

What do you think ?

Thanks
Thomas

Jeet
05-21-2006, 05:37 AM
i think multiple picks per person is good. say for example best 3

with priority

regards

jeet

ThomasAn.
05-21-2006, 08:01 AM
Hi all,

****Voting for this challenge has started******
Please feel free to make your choices !

Voting page:
http://mywebpages.comcast.net/doehtal-rx8/RenderingChallenge01/Round01-Entries.htm

Challenge Home page:
http://mywebpages.comcast.net/doehtal-rx8/RenderingChallenge01/ChallengeIntro.htm

-Thomas

YF19-sama
06-11-2006, 07:04 PM
Hi all,

Here is an entry from "hdesbois" and done with Maxwell.

I think it has great photorealism ! Quite impressive.
Thank you hdesbois for the entry !

Hi,
HAve he turned the top plane into emiter on his scene? I'm curious about the camera settings too because image stay sharp.

Even if the challenge is finish I show you my image http://www.deviantart.com/edit/34582428/edited

ThomasAn.
06-11-2006, 08:29 PM
Hi,
HAve he turned the top plane into emiter on his scene? I'm curious about the camera settings too because image stay sharp.

Even if the challenge is finish I show you my image http://www.deviantart.com/edit/34582428/edited

Hello YF19-sama,

A couple of points:
1. The top plane should not be an emitter.

2. I cannot see your image (says "forbitten" )


-Thomas

ThomasAn.
07-21-2006, 06:04 PM
Hi all,

sorry for the long absence.

A couple of things:
This challenge is ongoing indefinately. You can still submit images now or a year from now.
Every now and then the images will be grouped and will be posted for voting. This way we can look at the past voting rounds and see what we can do to improve and outdo the earlier tries.
Round01 of voting can be seen here (http://mywebpages.comcast.net/doehtal-rx8/RenderingChallenge01/ChallengeIntro.htm) (the link at the bottom of the introduction page).
Voting rounds are also open indefinately, you can still vote any time you want for present or past images
Please keep sumbitting images (please read the guidelines too)
Try to use the Nurbs files, because I see the OBJ and 3ds mesh files are coming up with some smoothing issues
Can you outdo the images from voting round 01 ?

ThomasAn.
08-12-2006, 08:22 AM
Hi all,

This is a preview of the upcomeing version 02 of this challenge. The scene contains more detailed high definition caustic elements.

This time the materials of the scene will be more explicitly specified to eliminate variablity.
It is is going to be a strict engine to engine challenge indicating the skill of the artist in their understanding and ability to handle the settings of their engine of choice.

Given the geometry and placement of the scene emitters, the scene sould have only one geometric optics solution. In other words if it was placed in a physics software package (like TracePro) the solution is unique... as is the case in real life ... so the expectation from all engines is that the result should be identical.


http://static.flickr.com/57/213025924_c67d6cfc1d_o.png

gent_k
08-12-2006, 04:07 PM
Hey, I saw this thread earlier but didn't take part. Saw the post mentioned again in a mailing list so I decided to participate this time. Did a quick first render just to check the lighting and FG, and if I have time I'll do one with higher settings (aa, caustic, glossies) of this scene or the next. Think some of the meshes came in corrupted, as can be seen on the reflections.
XSI / mentalRay

(see next post)

1:10 hours on a p4 3200mhz

gent_k
08-13-2006, 06:31 PM
OK, here's my entry. Had to remodel the rings, as the triangulated areas were screwing up the reflections.

XSI 5.11 / mentalRay
http://i97.photobucket.com/albums/l217/gent_k/jewelry.jpg
2:49 hours (p4 3200 mhz)

(was scared to turn accuracy of caustics up on my crappy pc :P)

ThomasAn.
08-14-2006, 02:28 AM
Hi Gent_k

Thank you for the image entry !!
Should I consider this one your final quality image ?


Thanks,
Thomas

P.S. Sorry about the mesh file and the bad geometry ... were you able to work with the obj NURBS file at all ? or the STEP file ?

gent_k
08-14-2006, 08:04 AM
Well yes, I don't plan to do another one, even though the caustics are quite noisy in this one.

About the files, I didn't even try the nurbs files as I was certain they wouldn't import correctly. XSI doesn't have that good support for nurbs and in any DCC app importing nurbs files is very tricky. For the next version of the scene I would say just go higher res on the meshes, as far as I see it efficiency isn't the main aim of this challenge.

Bao2
08-31-2006, 12:39 PM
Hey Thomas Anagnostou, I think the scale is not correct. Could you say
what is the measure of the top reflector in your scene file so I can check
that.
I will use mental ray and there the scale is important. This render behaviour
is different with different scales.
I scaled the 3ds file to 39,4% of original imported size and now my
top reflector measures: 311,04cm x 502,93cm and so I presume your
top reflector measures 311 x 503 perhaps?

That is 3meter x 5meter (and I scaled down to 39% the original imported
3ds file...). The rings are 41cm diameter aproximately...:)

Or perhaps I live in a modified hyperdimensional curvature in space time???
(yes I was reading these days very strange documents)

I wonder what the other people used, I was viewing no one cares of size
even using mental ray. A test I will do when post my attempt will be
scale again (scaling also the wattage of the lights, well damn lumens until
autodesk get concerned about that) and compare if the result is different
or not, I just now am not sure because the scene is very simple, no materials
involved so perhaps the scale is not a variable to have in mind. But I need
to test myself that because renderers are matematical formulas not real life.
I tend to think in real life this situation probably have the same result (with
perfectly glossy surfaces like proposed of course).

CGTalk Moderation
08-31-2006, 12:39 PM
This thread has been automatically closed as it remained inactive for 12 months. If you wish to continue the discussion, please create a new thread in the appropriate forum.