new workflow - some test renderings

[bernhard]

RWF - Reproduction

Hello friends !

Today I had to reproduce the look of a photograph I got from our client.
I thought, now it's time to test my RWF and let's see, if I have a chance to reproduce the same look.

@modeling: 45 minutes
@texture painting: 25minutes
@vray shaders: 15 minutes
@post: lens only for lights about 2 minutes
the renders are still low settings.....sorry for the strips, didn't have more time, and sorry, some more details are not modeled, needed to be done with lightspeed......

photo1:


repro1:


photo2:


repro2:


hmm, I think not to bad for the rieder work flow, so now I do have a test that shows it's possible to simulate different atmoshperes..... makes me happy ....

so far so good, wish you all the best and happy rendering for everybody,
bernhard

[paulison]

looks good for such a quick turnaround. it just needs a bit more detail and texture work but i think your goal of getting the lighting figured out is a success. Looks pretty close to the photo to me in terms of lighting and the rest would come good with more time.

was the client happy with the pots, statues and self promotion? :P

Thanks for sharing, I think most people develop a workflow and then keep it secret, stamp a TM on it and dont want others to know about it.

[bernhard]

thank you

Thank you paulison for your kind words.

you are right, I think with more modeling details and more time for textures and materials it would get very close.

@client:
lol: he will get some other shots, without my self promotion... hehe, it's just better than every watermark..... and I need to think to sculpture myself in 3d instead of using the statue or the teapots... lol:

@sharing:
you are welcome - would be great to see other testings too.....

take care,
bernhard

[jujubee]

Very nice. Basically you are using Linear Workflow with some minor modifications to the Color Mapping by using exponential. The other main component is that you've adjusted your camera for very high ISO settings - something which is not usually typical (nor would you do in real life.)

The renders are very clean - I am impressed by the whites, shadows, and the overall GI. Again - the majority of this is color mapping and film speed.

I do wonder about the following however:

1) how does it work on exterior images?
2) how does the accuracy of a color swatch (in real life) match up with the bitmap editor than translated to your final rendering?
3) how does the light multipliers work - do you have drastically various settings for multiple lights within a scene?

[bernhard]

it's RWF

@jujubee:
You meant, I basicly use the linear workflow... hmm, well, I don't think so, because if it would be linear, I wouldn't use 1.6 - that's the weired thing on my RWF , it's exactly the middle between 1 and 2.2

The film ISO is 200... I think a film ISO from 100 to 200 is fine, and is also used in real life.
I would say, the shutter speed wouldn't be as long, as it is... but I was talking to some studio photographers (pros) and they use also sometimes a longer shutter speed, and of course the cam needs to be on a tripod, would be shaky....

Majority of this is not alone the 1.6 gamma setting and some unusual extra high film ISO.
I would say, all of them play together so that you get this nice results.

Mainly, I set my focus to the levels... when I rendered an image, I always check the levels and colors in photoshop. If I don't reach the entire level range from 0-255, I know I have to change ... a good looking picture always shows you the full 8bit range from 0-255.... if not, there is something wrong with the light... in my opinion, that was a focus I had.... additional, I also always checked the textures in photoshop and matched it.

1) how does it work on exterior images?
yeah, very good idea... I didn't try the RWF for extirioir until yet, I always used the LWF.
But I will try.. it shouldn't be different, if I follow my rules it should work... theoretical, bt I need to make some tests of course...


2) color match
I really don't understand your specific question... but let me tell you, that color matching and using right textures for the render is very important.

if 2d color matching, or 3d color matching.... both is very important to get a photoreal look.
you can read my cg study about color matching and making your own hdris on my website here:

http://www.bernhardrieder.com/cgstudies.html

The key is here also the white and grey sphere, I used for matching the colors with the light environment, even if it's not 100% perfect, it makes a difference.





3) Light Mulitplier
That was also my goal, I wanted to avoid crazy light values, or extrem high multipliers... I figured out, that very strong and bright lights, caused problems in rendering, white spots appeared, gi caustics was an issue, and especially glass materials and refraction materials caused always problems with high settings...

So my idea was, to use regular standard light Values, measured in WATT. I never really could imagine, how much it is in real life if I used the regular mulitplier... 30, or 500 or what else... because it also depends how big your light source is (without scaling the light/Reset XFrom modifier) only using the accurate width and height settings... and so I use WATT now, because I know WATT from real life and can imagine...

so NO, I don't have drastically various settings for the lights, and the good thing is, you don't really need so much lights with the startup scene...

timmatron mentioned it before, it gives you really nice control when you start, and you can really manipulate the lights you would like to have, to get different atmospheres and looks you would like to achieve...


showed like above in the fast reconstruction of the light setup and look from the lobby...

ok, I hope that helps you,
take care,
bernhard

[jujubee]

Well part of LWF involves calibrating your monitor gamma - some monitors are calibrated at 1.6 (I think this was what Apple monitors were at - I can't recall precisely), 2.0, and others follow 2.2. It also changes depending on whether you use LCDs or CRTs.

The basic setup is LWF other than your color mapping. If you used Linear or Reinhard, or even Exponential - and have the gamma set to 2.2 with Max preferences set to what you have it as, the end result (final render) should be exactly the same unless Vlado has changed Vray since. At least that was the way it was.

If you push your gamma value down (Exponential to 1.6), you could potentially be properly calibrating your 'monitor' gamma towards proper linearization in accordance with your monitor - hence Linear Workflow.

Simply put, if your monitor (or OS like Apple) has a different starting gamma value and hasn't been adjusted to 2.2, shifting your color mapping to 1.6 may be linearizing your image anyways.

LWF ISN'T always 2.2 Gamma. People can make that mistake very easily.

Bouncing your ISO up would have similar results to rendering out to EXR/HDR and then changing exposure in Photoshop afterwards. The Vray camera value if course is not related to LWF - that's an independant image value.

When I get some time, I'll have to give your method a try and see what results I can come up with and if it changes drastically. Right now I'm too busy to drop what I am doing (setting-wise) mid-project.

The old forum has been deleted I think - I wrote a few tutorials simplifying LWF based on the work of Chris Nichols and (I'm having a memory shortage - someone help me - the cigarette commercial guy) before it became popular, and before other people tried simplifying it, then others simplified it, then others tried simplifying the simpled version, then videos were released. I'm not bragging but - I'm just saying I'm not a complete noob either. I really had to study intensely 'why' we were linearizing images to Max gamma and Vray - it was definitely complicated understanding the mechanics but well worth the time involved. It's a shame we lost those threads. I didn't come up with the method and can't claim to, but definitely made it easier in some ways for others to understand.

http://www.bernhardrieder.com/cgstudies.html

Your site is excellent. You have some really outstanding work there and tutorials.

I used to create HDRs with a camera - but this was back in 2000-01 when I had a state-of-the-art 3.2 megapixel camera lawl. My tripod was off camera center making spherical rotations and stitching hell - SLRs have since solved this problem. If I did do it again, I'd definitely go the route of using a fisheye lens and an EOS XT series. I also had this site: http://www.hdrsource.com but have since stopped making and selling HDRs digitally a couple years back.

[jujubee]

I'll add one more thing - Linearization (in regards to workflow) ensures that your colors across a broad range of applications are accurate.

However, LWF does not take into consideration Vray Lights or Max lights or even camera settings - those are completely independant. LWF basically took the old guesswork of choosing a color swatch in Max then when you rendered to Frame Buffer, it looked completely different.

All that LWF ensures is color fidelity if properly set up.

However, if you get a great render - that speaks for itself. Your renders are very good. Is it a new method or are you just properly calibrating? I don't know - I'll have to test it further.

[Morne]

Yes, this looks very good. Seems you understand your own workflow and VRay in general a lot better than me. I'm still trying to figure out LWF - LELE's tutorial and what others say and came up with in their renders, looks so easy and simple to understand, then why am I struggling so much. Anyway, that's another topic for a diffirent thread for another day.

Good job Bernhard!

[jujubee]

I did some tests. I didn't adjust my camera - it's left at default 'Video' for all of these. These are just test renderings using LC/LC at very low subdivisions - it's still a work in progress and the f-number is going to have to be tweaked. So excuse the 'final' quality - it isn't there yet.

There is a slight change in image difference between Linear and Exponential at 2.2:


Linear Workflow (linear 2.2) with settings above.


Exponential 2.2 with settings above.

Below is the gamma change of 1.6 (again, no camera adjustments):

Linear 1.6 with settings at top.


Exponential 1.6 with settings at top.

Aside from adjusting the camera values, I'd argue that Exponential 1.6 and 2.2 seems more washed out than Linear 2.2 and 1.6.

[tct70]

@jujubee

Can't see any images fella!

[jujubee]

Hmmm - Comcast is weird... Working now???

[tct70]

Sure is!

[bernhard]

hmm..

@jujubee:
great to see that you try thw rwf too. I am not really sure about your settings, your render looks a little bit pale... hmm, maybe because you have a lot of grey toned materials...

- did you use any vray light inside the building as well ?
- how's about a vray light in the window ?

maybe that will bring more color into your image.

@exterior:
If I do have time, I will try tomorrow an exterior with RWF, I'll let you know about the results.

testing, testing, testing... hehe,
bernhard

[jujubee]

Don't worry about the overall lighting of my image - it is shit and looks like shit. That was just a quick lc/lc test for client - I don't have much time to screw around with it to get proper lighting last night. I have since corrected it by simply raising my f-numbers but no time to post latest.

What I am comparing is your RWF to LWF and pointing out that your renderings look more washed out with your 'gamma' method. Your images (using exponential) are duller at both 1.6 and 2.2 minus camera adjustments. The only thing I didn't do was readjust light multipliers for all of them and camera settings. Compare the two images side by side. Your settings are actually grayer.

I will even add that with LWF, the highlights are even better (stronger.)

The only other thing you have done is adjust your camera ISO and film speed settings. Camera makes a HUGE difference here than the difference between 2.2 and 1.6 gamma (or LWF vs. RWF.)

[paulison]

Id actually question the whole point of what is the better method (I dont think there is one perfect way of working or even a better way) based on how we all have to adjust and tweak light and materials anyway until we are happy with them.

If your scene looks too dark with 1.6 gamma then you would probably compensate by increasing lights so that it looked good. Same if your scene was set up for 1.6 gamma, then to change to 2.2 gamma would mean that you would naturally adjust things in order to make your chosen workflow work.

So i hope this doesnt turn into an arguement about what method is better because that would be really pointless IMHO

I was more interested in the observation that the exponential tests looked a bit more washed out than the linear ones (because i havnt tested it that much myself)

Back onto the comparison in workflow. I've got one of those 24 inch Dell monitors and to be honest, after looking at the tests i think its gamma must be around 1.8 - i've tried calibrating it before and couldnt figure it out :P. almost all LWF 2.2 gamma images look too 'flouro' to evenly lit to me like someone has just changed the gamma in photoshop a bit too high - your typical client may observe the same thing with these DELL monitors.

Bernhard, i think you method works well for you and sharing this information may help others trying to achieve the same results. I also think that your HDRI tutorial is great and very thorough. If you have any other tutes you can write this would be much appreciated. I guess for me, I work as part of a production line and lose the motivation to test new methods and try new things so find that tutorials where other people have gone to this effort are very helpful.