[Radiance-general] Sunlight through glass

[webmaster at audice.com]

Just in case:

We have just completed a validation of 9 daylighting simulation
packages. This validation will be published in the reference book on
Lighting Engineering under the editorship of academician J.B.Aizenberg
and I can confirm that only Radiance and AGi32 provide correct
qualitative and quantitative results.

We've got 17.7% maximum relative difference with SuperLite 2 and
higher than 150% relative difference with Lightscape.

We wanted to test Inspirer from Integra Inc., Japan, because they
claim it provides phisycal accurate simulation results. But,
apparently, they didn't provide us with the package for evaluation.

RG> Jack de Valpine wrote:

>> Hi Giullio and others following this thread,
>>
>> <RANT MODE>
>> It seems to me that this is a problem with proprietary shrink wrap 
>> software renderers. Who really knows what is going on under the hood, 
>> whose word should be taken for the validity of the physical model, 
>> what validations have occurred?!
>>
>> Users see the product marketing materials that talk about Global 
>> Illumination and "physical accuracy," and they believe the hype! That 
>> seems to me to be pretty dangerous if you are going to be using the 
>> tool to somehow "validate" design issues!


RG> Can I get an amen-ahh!

RG> I got into it with a CAD/modeling pundit several years ago, when he 
RG> stated in a review that Viz was capable of producing renderings that 
RG> were "95% accurate".  What the hell does that mean?  Well, I asked him,
RG> and he said he spent a full day with Stu Feldman (one of the two 
RG> founders of Lightscape), and that was what he walked away with.  I told
RG> him he didn't get his money's worth on the half-day seminar.  We went 
RG> back and forth on this, I told him he was doing his readers a great 
RG> disservice, and in the end I was made to feel like a loser for being in
RG> such a small industry that no one cares about.  Something about "if you
RG> want to stick with Lightscape, good luck, it's not going to be supported
RG> anymore, blah blah..."  The good news is, it forced me to finally learn
RG> Radiance.  =)  The problem is that that review went into Cadalyst 
RG> magazine, which is read by lots of Architectural CAD drafters and 
RG> renderers, and as you mention they aren't necessarily concerned with the
RG> goings on under the hood.  And then you have people showing renderings
RG> to clients and saying "oh yes, that's what it will look like, we used 
RG> software that's 95% accurate!"  Which is amazing, since many of these 
RG> people driving the program don't know how to read a polar curve on a 
RG> luminaire cutsheet. Oops.

>>
>> I know that there are systems that have implemented GI to varying 
>> degrees and sophistication. But the problem is you probably have to be 
>> an uber expert to use them and/or code up custom material/lighting 
>> shaders. Still though the question is what validation has occurred. I 
>> think that most commercial renderers and users of said systems are 
>> really not that interested in physical validity, they are most 
>> interested in the outcome/appearance of the final image. It does not 
>> really matter how it get there.


RG> ...and if the result looks nice in half the time, screw accuracy. That
RG> was a lot of the complaining I saw on the M-R forums, was how slow it 
RG> was.  It's comments like that that really give you insight into how 
RG> little these people understand the problem of solving GI. 

>>
>> It seems to me that the one commercial product that showed some hope 
>> in its original (pre-acquisition) form was Lightscape. However, Rob 
>> Guglielmetti has explored and written pretty extensively on this topic 
>> seemingly with only partial satisfaction (my apologies to Rob G. for 
>> such a cursory summary, he did some really excellent work on this). 
>> Note I also believe that the original developers of Lightscape were 
>> truly interested in enabling people to use a tool with a reasonable 
>> and practical level of physical validity.
>> </RANT MODE>


RG> Thanks Jack.  I was merely trying understand the product's limitations
RG> because I wasn't ready to commit to learning radiance in production.  
RG> =8-)  And you are right about the developers.  At least in the case of
RG> Rod Recker (the other Lightscape developer), I definitely felt like he
RG> was interested in giving the world a tool.  Sure, getting rich off it is
RG> a nice bonus, but he seemed to be interested in taking that Cornell 
RG> education and channeling it into a lighting tool.  He's a really nice 
RG> guy and wish him the best of luck at Autodesk.  It's truly unfortunate
RG> what happened to that product, but then again it was severely limited in
RG> what it could do.  Of course, who knows what it'd be like today had the
RG> right people stayed in the decision-making seat.  Then again, how far 
RG> can one go with radiosity?  (Well, I guess the folks at Lighting 
RG> Analysts can answer that one, and the answer is "pretty damned far".  
RG> Their AGI product continues to improve, and with people like Martin at
RG> PSU doing validations with it, it becomes more and more robust.  But I
RG> still prefer Radiance.) 

RG> Interesting thread; sorry for the somewhat off-topic verbiage, but I 
RG> think we all learn from these exchanges too.

RG> - Rob Guglielmetti
RG> www.rumblestrip.org


RG> _______________________________________________
RG> Radiance-general mailing list
RG> Radiance-general at radiance-online.org
RG> http://www.radiance-online.org/mailman/listinfo/radiance-general




-- 
Best Regards,
Ilya A. Zimnovich                          mailto:webmaster at audice.com