Daz 3D is part of 
Connect
DAZ Productions, Inc.
224 S 200 W, Suite #250
Salt Lake City, UT 84101
Licensing Agreement | Terms of Service | Privacy Policy | EULA
© 2024 Daz Productions Inc. All Rights Reserved.Daz 3D is part of
Connect
DAZ Productions, Inc.
224 S 200 W, Suite #250
Salt Lake City, UT 84101
Licensing Agreement | Terms of Service | Privacy Policy | EULA
© 2024 Daz Productions Inc. All Rights Reserved.
Comments
Here's a test render where I changed the light color to blue, but kept it very dark as in Rashad's original scene.
I also turned off two of the reflected sunlight replicators, and reduced the number of replications in the visible replicators. This decreased render time, yet kept the diffuse shadows. The scene is bit darker because of the turned of replicators and reduced replications.
I do prefer the more bluish reflected light, which is closer to the reflected light in the Octane and GI renders Rashad posted as examples. The other effects on the render due to my changes won't be as close.
It looks great! I can only see one thing to consider altering.
It looks like your 3D Fill has shrunken a bit causing your light to collect at the center of the room a little too much. I should have mentioned earlier that the exact scaling of the 3D Fill object is essential to maintain as changes to the density of the lights will alter the size of the Fill itself. You removed lights so the 3D Fill shrunk a bit. Not enough indirect light is reaching the edges of the room.
If you have the time, I suggest re-rendering the scene after you restore the 3D Fill to its original scale.
You're right! I forgot to adjust the grid size to compensate for the lower number of lights. As Homer (as in Simpson) would say, D'oh!
BTW, I'm re-rendering the scene with lights set to a brighter blue color and the light intensity set to .35, and it seems to be working. The field where you enter the value says 0. If I recall when I tested this, if you enter a value greater than .50 it will display as 1.
So, this one uses a brighter color blue for the light, and it is set to .35 intensity. I went back to the old number of replications and as a test turned off another bounced sunlight replicator.
I could probably turn down the intensity of the light a bit more and perhaps adjust the blue light a bit more towards gray, but it proves that you can enter decimals into the intensity field of your lights and Carrara will respect them, even if it doesn't display it correctly.
Interesting find Evil, I never knew that we could enter decimal amounts lower than 1% into that field, the render looks great, I'm really excited about the possibilities of this approach :)
What was the render time like?
Waiting with abated breath for Episode #3... :) :)
That was just a reference to this statement I've read a few places recently. Here's an example.
http://en.wikipedia.org/wiki/Unbiased_rendering
From Wikipedia, the free encyclopedia
"In computer graphics, unbiased rendering refers to a rendering technique that does not introduce any systematic error, or bias, into the radiance approximation. Because of this, it is often used to generate the reference image to which other rendering techniques are compared."
I find a lot of folks tend to judge "realism" of a render using nothing more than a "looks good to me" evaluation. But unless you can use a really physically accurate image (ie, a photograph) then you are judging a fake using another fake.
And in the years I've been in this forum, I can't recall ONCE anyone comparing their renders with a photograph.
Just my 2 cents... :) :)
I absolutely agree. Real photos are most certainly the preferred approach. But there re also some major caveats. For a real photo to have maximal value it requires a lot of information about the conditions surrounding the original image, such as the type of lens, the type of film, the exposure length.
The "What Color is this Dress" debate is a perfect example of why a real photograph can be extremely misleading and can send one down the wrong path. Only once one understands the conditions of the original photo session sch as realizing that the dress image is extremely overexposed and shot with a crappy phone camera does the reality of it's blue black nature take shape. But most people, judging only by the overexposed image, would be trying to recreate what they were seeing which would have been a white and gold colored dress. Just plain wrong.
So while I can concede real photos are extremely valuable, I also realize that they are limited in value and can even be misleading.
If I could have found a real photo of the scenario in questions I would certainly have used it. By using the Octane Render as a reference (which is what the professionals would do) gives us a close enough approximation to feel confident about the value of the testing results.
Interesting find Evil, I never knew that we could enter decimal amounts lower than 1% into that field, the render looks great, I'm really excited about the possibilities of this approach :)
What was the render time like?
Thanks Jon.
The render took less time than a GI scene would have taken. I didn't look at the final time. I doubt it would have been a good benchmark anyway, as I didn't render it over my network, and my main machine is a single processor G5, so it's pretty ancient. I would hazard a guess at around twenty minutes or so.
Thanks Jon.
The render took less time than a GI scene would have taken. I didn't look at the final time. I doubt it would have been a good benchmark anyway, as I didn't render it over my network, and my main machine is a single processor G5, so it's pretty ancient. I would hazard a guess at around twenty minutes or so.
Jon, You should download the scene file and test it yourself. I'd be very curious to know how quickly it renders for you
It should render in roughly half the time a full GI would have taken on your system. In the case of Evil having removed even more lights the render time could be only 1/3 that of a full GI render. You end up with an acceptable quality with a very minimal render time by comparison.
I found that the standard response in these here parts to a suggestion that an actual photo is the best reference for "realism" is just that...a lengthy explanation of the limitations of a typical photo.
And that's fine. But in the scheme of things, you only have a few choices if you want to gauge how realistic your renders are.
A. You can use your own personal "looks good to me" evaluation, which, in most cases, is severely lacking, for many reasons.
B. Or you can use an unbiased render, which, in most cases, is severely lacking because the artist never used any meaningful guide for his final product, other than A above. :) :)
C. Or you can use a photo. Which, if done well, gives the best, documented, reproduction of real life that you can find and share with others.
Not many other choices.
And you can easily get past the limitations of photos by using the following procedure:
Get a camera. Note the camera properties (focal length, etc.). Go outside. Take a photo. Note the sky conditions, and other lighting conditions. Take detailed photos of all the textures/surfaces in your image. Go inside and try to reproduce the exact foto in Carrara.
The point is, if you KNOW the conditions when you took the photo, there is no excuse for not being able to try to duplicate it.
Now, I realize that many people here like to play with computer graphics just for the fun of it, and an exercise like this is too much like school work, but I think it's a valuable exercise for those who REALLY want to reproduce realism, and don't want some easy, push-button solutions like unbiased renders or gamma= 2.2 :) :) :)
I think you might be misunderstanding what this Wikipedia entry was saying or implying. Yes, *IF* a professional, who fully understands all the complicated and detailed components that go into making a truly realistic scene (lighting, surfaces, etc.), needs to generate the most realistic images possible, they will likely use an unbiased renderer since it does not use the shortcuts that other renderers might use. It provides the most physically based rendering techniques.
However, it's like the Gamma=2.2 discussion. You first need to understand what's really behind it, not just click a button and pronounce the resulting image "realistic".
I think it's a mistake to imply that since professionals use it as a reference image then everyone can automatically use it as a reference image. Because that's clearly not the case.
And then as Rashad pointed out, sometimes a reference photo is unavailable or impractical. It seems that while you claim folks here easily dismiss your reference photo teachings, it also seems as if you dismiss what they have to say and take umbrage when the point is repeated.
Or did I misinterpret your comment that you were waiting for episode three with abated breath?
I'm not taking umbrage any more than you or Rashad are taking umbrage for me stating my views. Partly because I don't know what an "umbrage" is :) :) :)
Of course, sometimes reference photos are unavailable or impractical. Nothing is perfect. But in the scheme of things, I think you'll agree, that it is usually the last thing people will try when going for realism here, for reasons I don't really understand. Maybe you do. Not sure I'll ever understand why the suggestion always results in excuses why reference photos are not a perfect idea. But, all I can do is suggest, and if people aren't interested then they aren't interested. It's just a continuous head scratcher for me. :) :) :)
Especially since, in one breath, people espouse how "professionals" use unbiased renderers as reference images, but in the same paragraph dismiss one of the most often used techniques by professionals trying to go for realism, and that is to use photos. Even the camera mapping discussion was a clear example of how professionals use photo images to generate realistic 3D renders.
I'm not taking umbrage any more than you or Rashad are taking umbrage for me stating my views. Partly because I don't know what an "umbrage" is :) :) :)
Of course, sometimes reference photos are unavailable or impractical. Nothing is perfect. But in the scheme of things, I think you'll agree, that it is usually the last thing people will try when going for realism here, for reasons I don't really understand. Maybe you do. Not sure I'll ever understand why the suggestion always results in excuses why reference photos are not a perfect idea. But, all I can do is suggest, and if people aren't interested then they aren't interested. It's just a continuous head scratcher for me. :) :) :)
Especially since, in one breath, people espouse how "professionals" use unbiased renderers as reference images, but in the same paragraph dismiss one of the most often used techniques by professionals trying to go for realism, and that is to use photos. Even the camera mapping discussion was a clear example of how professionals use photo images to generate realistic 3D renders.
Rashad said that in absence of a good reference photo the pros would use an un-biased render. He didn't say that was the preferred method.
You keep saying that the last thing people do here is use reference photos, which you don't understand. I say, maybe you need to reevaluate that belief. I know that I have read numerous posts by individuals here, mentioning that they used reference photos and images. Probably more than half the Carrara Challenge WIP threads have reference images or photos that someone has posted at one point or another.
Yeah, I don't want to say you misinterpreted cuz I know you tend to take umbrage at anything that is said here that isn't positive...
But yeah, I think you misinterpreted. Rashad is a smart guy, and has some very good ideas and input. I like seeing smart guys give good input here to balance the often misguided expertise that gets promoted.
So yeah, I'm looking forward to seeing his input. Especially with the #1, #2, #3, teasers he posted... :) :) :)
Yeah, I don't want to say you misinterpreted cuz I know you tend to take umbrage at anything that is said here that isn't positive...
But yeah, I think you misinterpreted. Rashad is a smart guy, and has some very good ideas and input. I like seeing smart guys give good input here to balance the often misguided expertise that gets promoted.
So yeah, I'm looking forward to seeing his input. Especially with the #1, #2, #3, teasers he posted... :) :) :)
When you say misguided, you mean like, when somebody says nobody here uses reference images because it's like school work? :) :) :)
Rashad, I'm having trouble getting the file downloaded to play with. Is there some trick to download it? Evil, I know you got it already so it must be something I'm doing wrong. Are there steps I should be following? It brings me to a page where I can see the .car file, but can't seem to download it...
Click here
REALLY ?? Wow, I'm shocked. In all my years visiting here I can't recall the last time I saw someone take a real life photo and try to match it for realism in Carrara. Although I do recall a few years back maybe I posted a very simple photo of a bollard and suggested people try to match it, and the response was a fairly uniform "why don't you do it for us if you're so smart" or something along those lines... :) :) :)
Unfortunately I don't have the inclination to sort through all the old WIP threads, and since I generally ignore them, I'm unaware of any cases where people try to match photos for realism in those threads.
Do you have any specific examples of people trying to match for realism with a photo here? I'd gladly change my view if I'm all wet on this.
Thanks Rashad, I got it now (not sure how I missed that the first time :) ) I'll have to play around with it after dinner, but looking forward to it.
Visually this result looks great. It is much brighter however than either the Octane or the Carrara GI samples. The level of brightness you've given the 3D Fill seems more appropriate once the ceiling spotlights are turned on and there is more light expected to bounce further around the room.
There should only be three total sunlight bouncers, two exterior and one interior. The only way to actually turn them off I find is to disable the visibility of the light source itself, not its replicator. You are still getting a nice amount of upward bounced light so I'm curious how you've gone about the light disabling.
Generally, this 3D Fill approach as you can see provides a great deal of flexibility, especially in the way the colors are assigned. Fun fun
REALLY ?? Wow, I'm shocked. In all my years visiting here I can't recall the last time I saw someone take a real life photo and try to match it for realism in Carrara. Although I do recall a few years back maybe I posted a very simple photo of a bollard and suggested people try to match it, and the response was a fairly uniform "why don't you do it for us if you're so smart" or something along those lines... :) :) :)
Unfortunately I don't have the inclination to sort through all the old WIP threads, and since I generally ignore them, I'm unaware of any cases where people try to match photos for realism in those threads.
Do you have any specific examples of people trying to match for realism with a photo here? I'd gladly change my view if I'm all wet on this.
I didn't say they tried to copy the photos or images. I said they used them as references. If you want a copy or duplication that is different, and I can't point directly to one. But a reference is different than a copy. A copy doesn't really show off any artistic skill, especially if it is a copy of one of the quadrillion images on Google taken by somebody else. It is a technical exercise only. Not that that can't be useful. but that is not just what the challenges are about.
Since you don't have time to do a render here, I don't have time to sift through the challenge WIP threads, so here's the link to all the WIP threads up to this point:
http://www.daz3d.com/forums/discussion/45254/
I remember your challenge. It was a post or something wasn't it? I didn't try it for a couple reasons. The main reason was that I felt the image was dull. If it's not something that I would look twice at, why would I spend hours trying to copy it.
Rashad's room is another matter. It's still a dull room, but I don't have to build the set. With your post, I would have had to build the post and the cobblestone or paved surface that it sat on. Here, we're all working on the same scene file (like a control as it were) and just tweaking the lighting. The goal isn't to make a lamp that geometrically doesn't look "computery," it all does. The thing that is important in this exercise is the lighting. I think if this thread progresses, shaders will need to be looked at as well, since lighting and shading go hand in hand.
Visually this result looks great. It is much brighter however than either the Octane or the Carrara GI samples. The level of brightness you've given the 3D Fill seems more appropriate once the ceiling spotlights are turned on and there is more light expected to bounce further around the room.
There should only be three total sunlight bouncers, two exterior and one interior. The only way to actually turn them off I find is to disable the visibility of the light source itself, not its replicator. You are still getting a nice amount of upward bounced light so I'm curious how you've gone about the light disabling.
Generally, this 3D Fill approach as you can see provides a great deal of flexibility, especially in the way the colors are assigned. Fun fun
I may have just hid the replicator on the one by the window. Weird. my results have usually been the opposite, in that if I hide the replicator, all the lighting (except for the original light) is hidden. I did have a couple documents open while I was messing about with your scene, so maybe I thought I had turned one off and didn't. I didn't save that iteration, so I can't check it.
I shall have to experiment tomorrow.
REALLY ?? Wow, I'm shocked. In all my years visiting here I can't recall the last time I saw someone take a real life photo and try to match it for realism in Carrara. Although I do recall a few years back maybe I posted a very simple photo of a bollard and suggested people try to match it, and the response was a fairly uniform "why don't you do it for us if you're so smart" or something along those lines... :) :) :)
Unfortunately I don't have the inclination to sort through all the old WIP threads, and since I generally ignore them, I'm unaware of any cases where people try to match photos for realism in those threads.
Do you have any specific examples of people trying to match for realism with a photo here? I'd gladly change my view if I'm all wet on this.
I don't want to travel too far down this rabbit hole, but I wanted to add a couple of things.
Joe,
If I'm not mistaken it was StringTheory's attempt at replication of a real world photograph that lead us to this discussion in the first place. This was from a post made in the Carrara render thread just a couple of days ago:
http://www.daz3d.com/forums/discussion/43528/P615/#766924
I'm sure there are a thousand other examples of people using real world photo references. And I would venture to guess that many of those studies ended in nothing but frustration because the user discovers that no matter how much tweaking he/she isn't able to reproduce results that look anything similar to the original photo.
Edit:
I think I remember several years ago someone posted an image of a dock of some sort. There was a pylon or something like that. I remember lots of discussion as to why the pylon was shaped as it was in the original photo. It took a lot of guessing from forum members about the angle of the sun and the angle of the camera to get the pylon to taper at the end the way it did in the photo.
The study was insightful, but I'm not sure how many lessons learned from that experience have been translatable to later scenarios.
But it was interesting, no doubt!
yes it was here http://www.daz3d.com/forums/discussion/25644/P180/#393718 but the discussion got quite err... rambunctious
Anyway, reality isn't all it's cracked up to be - it's an optical illusion brought on by an absence of imagination (or alcohol) :)
oh good, I thought I was missing something :) I'm still waiting for a render that shows the renderer's blurry nose in the lower middle and eyebrows in the upper periphery, not to mention eyelash tips diffracting (/?) light. And then we are Totally missing Entoptic phenomenon http://en.wikipedia.org/wiki/Entoptic_phenomenon and I guess http://en.wikipedia.org/wiki/Optical_phenomena. Sorry to off topic :)
IF people want to learn about the real world and how to generate realistic images, there is a HUGE amount to be learned by trying to understand and duplicate a real world photo. You can start out simple, like the bollard I suggested years ago, or more complicated. It is not merely a technical exercise, it is a way to learn stuff. Yes, it can be frustrating, but if that is a deterrent to trying then you'll never learn anything.
But clearly it's not something the majority of folks here are interested in. Yes, there are scattered examples of people trying to replicate (aka, learn about) real world lights and textures and modelling by trying to replicate or expand upon a real world photo. But after any years on this forum I can assure you it's not something that most people have any interest in.
Nothing wrong with that whatsoever, I just think there's a benefit to mentioning it as a tool when people propose alternate views. No need to get upset or take it personally.
Ah, yes. How eerily similar the discussions of today are. It's a complete deja vu line for line, Hilarious!!!