A render using LWF

[rmejia]

I do agree with others that LWF images might look rather "dull", but that can be fixed with 1 minute of photoshop. Only thing is the colors change slightly. I personally think that even the "dull" straight out of the VFB images look alright. Using gamma 1.0 I think they look pretty bad and weird, shadows come out black and colors look to saturated.

I don't know if andronikos did any post work, but his images look perfect.

[dlparisi]

We have used neutral lights in the past, but there is a quality to the tonal balance and depth that the vray sun/sky provides that you do not get with neutral lighting. We even approached a method of using a blueish environment light and neutralizing the value in direct sunlight through the RGB's and intensities of the lights. Nearly impossible to push this method to an acceptable accuracy.

Pushing and pulling the physical settings puts it as close to realistic as we might get with swatch to print color accuracy. As flat as it may look when compared to the more stylized images seen here, it would be much flatter with neutral lighting.

We might not have the best looking images here on the forum, but they are straightforward, and we are not afraid to have a client hold a manufacturer swatch to our print. It has been quite some time since we have had to come up with any excuses as to why our colors did not look almost exactly like the samples that the client had picked and provided, and at least I have the peace of mind to know how to manipulate the vray sun/sky to suit the rare occasion that I can provide a more "realistic" set of images.

But where are you measuring the accuracy of the colors in the image? I sampled your image in photoshop for the stucco colors and there's a pretty wide range of intensities, even in just the areas directly lit by the sun. With all of the bounced light going around how can you back up the "1% deviance" number. And I don't understand your comment about the "quality to the tonal balance and depth that the vray sun/sky provides that you do not get with neutral lighting". Isn't this tonal balance and depth exactly the stuff that screws up your scanned swatches to something that doesn't match.

I can feel your pain about matching a clients swatch though. It's just that I think you have done your clients a disservice by fooling them into thinking that the swatch they're holding in their hand will look the same on the building in full sunlight. Your image colors might be "accurate" but your image also doesn't look like it's being lit by sunlight. If the intent of the images is to show a client color layouts, design,etc. they're fine but I wouldn't call them photoreal.

[beestee]

Minus texture, which we cannot get around(think our images are flat now?), we sample a smooth surface material under these conditions and check for value deviance in R, G, and B independantly...1% of the standard RGB 256 value range is about 2-3 points, which is much better than acceptable.

Believe me, we have played these cards before, and attempted to explain that showing a semi-realistic image with swatch accurate colors in full-lit condition is deceptive. We even rendered out examples to match existing condition photographs with renders that matched exactly to prove our case...but as strong as our case was, it still came down to the fact that these images are being viewed by various city councils under fluoroscent lighting condition.

Fact of it is, they can take our print outside if they want to see what the colors will look like outside, just as they would a swatch or material sample, so in a way, it is no more deceptive than your exemplary photographic rendering...or perhaps we should provide both so that they do not have to go outside?

[Neilg]

Oh - and yeah - he mentioned this was for animation so I think there was very little post-processing (if any) on his renderings here.

Where on earth have you got the idea that it's impossible to make an image look like that without LWF?

A workflow adjustment isnt a good benchmark to what renders are capable of, either.

[dlparisi]

Minus texture, which we cannot get around(think our images are flat now?), we sample a smooth surface material under these conditions and check for value deviance in R, G, and B independantly...1% of the standard RGB 256 value range is about 2-3 points, which is much better than acceptable.

What I also meant was what about the orientation of the surfaces. Surely a horizontal surface will get more light on it as well as other vertical surfaces might get less light. So is it just the one vertical face that you are trying to match?

Believe me, we have played these cards before, and attempted to explain that showing a semi-realistic image with swatch accurate colors in full-lit condition is deceptive. We even rendered out examples to match existing condition photographs with renders that matched exactly to prove our case...but as strong as our case was, it still came down to the fact that these images are being viewed by various city councils under fluoroscent lighting condition.

Like I said, I feel your pain on this. Funny how though if you were to show them an actual photograph of the building they'd be fine with it vs. if you tell them it's a rendering they're suddenly saying "It doesn't match!" Damn board meetings !

Fact of it is, they can take our print outside if they want to see what the colors will look like outside, just as they would a swatch or material sample

I don't think it works like this . Maybe they should hold the board meeting outside ?

[werticus]

If I have a client mention colours being off I just ask them about 300 questions as to what conditions they are viewing the swatches to compare to my images...

thing like: what is the colour temperature of your office lighting, and how many lumens it it outputting, also what type is it.

what is the IOR and material of your glasses if your wearing any

what angel are you holding the swatch on

how colour blind are you

what is your monitor type, if LCD, what type of lcd, what is it calibrated to, what's its variance.

etc

etc

and then ill explain that the paint maker does not guarantee the sample will match the paint also

9 out of ten times the client has a junk LCD with just f4cked colours. And a print from me resolves the issue.

all i do to match colours is use lele's .5 method for texturing and colour correct to 1.6 gamma and the colours match up perfectly to the paint makers specs as far as any reasonable eye can tell, assuming one factors in the various lighting conditions anyway.

[rmejia]

If I have a client mention colours being off I just ask them about 300 questions as to what conditions they are viewing the swatches to compare to my images...

lol, does he ever ask you again about the colors being off ?

[jujubee]

Isn't this tonal balance and depth exactly the stuff that screws up your scanned swatches to something that doesn't match.

I've said this before - Beestee disagreed to a point. I think exteriors do not work well currently with LWF. I think something Vray (in particular the Vraysky gradient) and indirect lighting doesn't contribute enough when working in this mode.

I have yet to find a realistic looking LWF method for exteriors. I just can't get it to work/look realistic.

Where on earth have you got the idea that it's impossible to make an image look like that without LWF?

The first image is a really bad example - no offense.

While it IS possible to get results similar to the second sets of images, it often involves a ton of post work for which I think Andro did very little seeing that it is for animation. Try getting those results for a still - now try getting those results for an animation. There's going to be a huge difference in the amount of post processing to achieve that quality with any other method other than LWF.

Honestly speaking, his results (Andros) look exceptionally realistic. You don't get that too many times from animations - either coming from big firms or small ones.

A workflow adjustment isnt a good benchmark to what renders are capable of, either.

This is true. Ultimately, if your renderings look like a cartoon and you're striving for photorealism, then you have a lot of work to do. There's no 'right' or 'wrong' method - but understanding that Max is fucked to begin with and that's why people use LWF (cause mathematically it's correct) is the concept behind all that. It's 'more' correct and it has its benefits which a lot of people fail to see.


All this aside...:

In all reality, I don't care what method other people use. If a person's renders look like crap, then that simply means more work for me. If a person's renders look good, congratulations - how much money did you spend on hardware, how many people did you throw at that job, or better yet - how many months did you spend working on/perfecting one single still? If none of those apply and it looks consistently great - please teach me master.

The proof is right there - if a person's images look like crap, then a person shouldn't give advice but really only ask questions. I hate to be mean about it. I've got some really good results recently, but haven't had a chance to update my portfolio/website in about a year. I'm also not here to 'prove' myself but rather to contribute.

I'm getting tired of arguing LWF or not LWF - I try to help people out and the only people that seem to argue are the people that don't know how to use it in the first place.

P.S. - this rant was directed towards anyone in particular. I'm not trying to single anyone out.

[beestee]

What I also meant was what about the orientation of the surfaces. Surely a horizontal surface will get more light on it as well as other vertical surfaces might get less light. So is it just the one vertical face that you are trying to match?

This is where we invite our old friend photoshop to the project...no room for any lively rogue colors

[beestee]

The first image is a really bad example - no offense.

None taken. I am not too proud of the bulk of our work, but it is proof that there is room for exterior LWF in production when using the vray sun/sky. Let me see if I can dig up some of our work for different clients that use sun/sky and LWF on exteriors, maybe they will have a little more sway...

Keep in mind that these are the "raw" renders. Back when they were done, they did have some corrections in post before they were sent to the client, but that is not what needs to be shown here:

[jujubee]

The last image looks the best but the default Vraysky is all messed up - tha'ts what I've been saying - it's hard to get a good balance with LWF there.

For now the only solution I've been able to come up with is to replace the sky entirely.

[beestee]

We are basically screwing up the way the VRay sky is calculated by using LWF, and this particular map cannot be corrected by simply placing it under a colorcorrect map. It is not just the appearance of the sky that loses fidelity, but every aspect in which it is used to add to the rendering, including the lighting. So even though it can be made to work, makeshift practice probably isn't good to advise, so I do see a little more clearly where you are coming from

Hopefully we will see this problem addressed eventually...although the proper route for that fix (autodesk) may not even see it as a problem.

[Morne]

it seems to me more and more that LWF is NOT the way to go and that it creates more problems than it solves...

[jujubee]

I find it to be a problem with exteriors using Vraysky - yes. But this is totally NOT the case for interiors.

[dlparisi]

it seems to me more and more that LWF is NOT the way to go and that it creates more problems than it solves...

I disagree. I use LWF and vray sun/sky/camera almost daily for exteriors and I can quite often just use the vray sky for the final image rather than replacing it with a bitmap. While the attached image isn't the greatest (it's a WIP), the sky looks failry well balanced to the facades. Exposure on the vray cam was f-16, 1/100s, 200ISO. A quick google image search of buildings and skies also yields skies that are much darker than you may expect. I don't expect the vraysky to visually match a photgraphed sky though as the vray sky is an ideal, mathematical model of the sun so it'll lack some of the elements (clouds, haze, etc) that make it looks totally convincing though. Of course for an interior it quickly blows out but we're talking just exteriors here.

IMO, using the VRaysky with LWF actually works better than without it. My 2 cents.