[Half Life]

I'm doing a bunch of research on the settings for the "physical sky" settings in Maxwell trying to get the most realistic settings for any given particular location on earth -- I've come across some interesting webpages that I think may be useful:

Todays Ozone Map:
http://toms.gsfc.nasa.gov/ozone/ozone_v8.html

Todays Aerosols Map (although I'm not sure if these are based on Angstroms turbidity or not):
http://toms.gsfc.nasa.gov/aerosols/today_aero_v8.html

Real Time water vapor map of the USA:
http://www.gst.ucar.edu/gpsrg/realtime.html

Planetary Albedo Map:
http://earthobservatory.nasa.gov/IOTD/view.php?id=2599

Hopefully I can get to the bottom of some sort of system that will yield locally realistic results for the atmosphere (minus clouds of course).

Any input is more than welcome.

Best,
Jason.

[Richard]

Mate I found a stack of these in image form somewhere years ago - mind you "without clouds" it's never going to be realistic!

[Half Life]

Hey Richard, I agree and disagree... I think clouds (along with moon and stars) would be an awesome addition to the realism of the Physical Sky, but I think that if you can get the local atmospheric settings to align with the Physical Location and time/date then it would be much easier to do several things:

1) Match Photos based on atmospheric conditions at the date/time/location of the photo.

2) Visualize how an object will look in a particular area on a clear day at any time of year.

This would be particularly useful in the Arkviz sector I would think... but may also be useful in film as well.

I would love for that information to be somehow automatically fed into the sky parameters when you pick a location/date/time but I think that is probably pie-in-the-sky thinking right now.

Best,
Jason.

[Richard]

Yeah mate I understand what you are saying and in the aim of true results it would certainly pay, that said for testing yes BUT for arch viz for the most maybe not so! A sky and particularly clouds can really be used for added dramatasism (sp?) - particularly where clouds can be used direction wise to draw out a dramatic roof silhouette!

Useable volumetrics are possibly the most critical in assuming a true physical response and again without them one must resort to post work.

The reason I myself dug deep years ago to collect this information was for the very reasons you suggested, yet after initial use I dropped it! It is evident that most arch viz wants for drama in their scene and why most scenes are shot with a sunrize of sunset theme!

[Half Life]

Obviously in that situation HDR skies seem to be the smart choice -- although you do run into the issue of repetitive looking skies after a while. HDR seems like a stopgap measure to me for a few other reasons as well...

I've seen a few Sky simulators out there like Ozone (http://www.e-onsoftware.com/products/ozone/ozone_4.0/) and Stellarium (http://stellarium.org/) as well as several others.

None of them really deliver what we need, but I do have longterm hope from an unlikely ally -- Flight simulation... check this out http://www2.hifisim.com/node/1044. I guess they have a need for "true-to-life" weather for the games but all that information could be re-purposed for feeding data into a more dynamic Physical Sky at some point in the future.

Best,
Jason.

[Richard]

Wow that ozone app is cool darn shame they don't have a PS plugin! You are right that HDRI has it's limitations, particularly given the quality needed for generating shadows, and hence why I requested ages ago for the option to mix MR's physical sky sun with the IBL illumination channel to be able to control this aspect better.

[gotoxy av-media]

have you tried to setup your sky/clouds with ozone and then render a 360°pano as an hdri!

munch

[gotoxy av-media]

a small test using vue...

http://rapidshare.com/files/441269903/P ... _pano6.zip


munch

[Half Life]

That is certainly an option and for certain types of subject matter it may in fact be the very best workflow -- my thoughts is it is probably not the best workflow when you want to place your model in the "real world" meaning atmospheric conditions that exists (or has at some point in the recent past) at an actual point on the globe.

We have non-stop weather channels feeding vast amounts of information about the atmospheric conditions in near real-time covering a very large portion of the globe -- it would be great to be able to tap into all that information so that you could either look up a particular longitude/latitude or simply a city name and get precisely the information you need for physical sky settings for that location, time, and date.

It would be even better if that could simply be fed into Maxwell's Physical Sky system when you choose a location, date and time (but that would have to go on the wish list)... and since we are dealing with wishful thinking, the ability to allow the dynamically loaded settings for the "Physical Sky" to be "animated" over time would be awesome -- capturing a real "day in the life" would be a pretty cool presentation tool... or even going through a year in the life (seasonal atmosphere changes).

The lack of clouds, stars, and moon are issues that are secondary to me (I've already voted for them in the wish list forum) -- my primary goal is to find a way to use the tools we currently do have to the fullest, best, and most accurate possible level.

Best,
Jason.

[brodie_geers]

I think Richard had a good point though. I think there needs to be a give and take there somewhere. In my mind, that's where the current environmental system, particularly the GPS globe, becomes less useful in most circumstances. I suspect that there are some instances where someone might want to find out what a building looks like, not only on a particular part of the planet, but also in different atmospheric conditions...or even more likely, different seasons. But although, I think that ability should be there, I think it must come secondary to being able to use artistic license to position the sun and clouds because for 99% of us 99% of the time, our clients just want a rendering that looks realistic and best shows off the product (even if that means a north facing wall is actually facing south in the rendering or some such thing).

I guess it's all dreaming right now at any rate, but things are certainly heading in the direction you're talking about so I guess it's just a matter of time.

-Brodie

[gotoxy av-media]

hmm, ok!

maybe i don`t understand what you are looking for...
you want to feed in data like location / date / atmosphere etc. of the real world?

i think this wish will never come true, even if you can use these data - it is only a rough approximation -
i prefer faking my hdri-env until it fits to my needs!


munch

[Half Life]

Yeah, artistically I'm with you -- but Maxwell is a physics-based simulator first and foremost and simulations need accurate data to be realistic... I'm just trying to get to the bottom of how to get the most accurate data possible.

Artistic concerns are less important at this point, but would be sure to benefit from advances on the simulation front (as always).

Vue is a fantastic program if your primary concerns are aesthetic or artistic... but lacks a bit in realism due to it's bias toward the aesthetic and artistic.


Best,
Jason.

[brodie_geers]

I'm totally with you, it's really a discussion for another day. The first step is, of course, to get the realistic environmental parameters (both the numbers, as you're doing as well as the additional factors, ie. clouds, that are yet to be incorporated).

-Brodie

[Half Life]

As far as reality based atmosphere (and cloud) information this website is a wealth of free information:

http://modis-atmos.gsfc.nasa.gov/

This is the type of data that could go a long way towards the type of physical sky that would be much more realistic for any given location, time, and date.

Best,
Jason.

[Richard]

Mate I had a long think about this the over the weekend whilst sitting at the beach and looking around at the immediate local environment.

Firstly I'm not pooh poohing the suggestion you are making! Though it became obvious there are SO many more environmental aspects that would effect any attempt at a physically correct result and it is I would think way beyond scope to collect enough data to achieve such.

We discussed clouds and given you are assuming a physically correct output for a certain day, time etc cloud cover for that day, minute and second would be possibly the most critical.

Also how far from a surf beach, how big was the surf, what height are the coastal dunes, what elevation is the site above the ocean, the wind direction on the day, the amount of hard stand around the site, the height of surrounding hills are the slopes facing the site in shadow, what type of trees are predominant around the site - do they give of light effecting vapour, what was the local temp? etc etc.

The only reason I could consider the need for a true physical response was in the case of an application for the development going to court for final determination - then to suggest in anyway that your render was a "physically correct response" would place you at the will of an opposing expert witness. A null answer to any of these local environmental aspects would destroy your credibility and hence place your results as questionable. Better I think to photograph the background on that day and aim to match lighting as near possible to that evident and state it "as approximate" - then no questions!