Self-Inflicted Challenge

[Jake]

There was another discussion at the SketchUp forum asking the eternal question, "What is the best renderer for SketchUp?"

I've read that about 100 times, so I thought I'd try a little self-inflicted challenge to produce an image similar to one posted at the SketchUp Forum.

http://www.sketchup.com/forum/download.php/3,7024/bagno.jpg

The object image was done by Valerio using V-Ray. His work is outstanding, so I'm using it as a goal image. There is also a lot of talk about the Maxwell. I believe Cheetah3D has the potential to get similar results as both of these programs.

The image I've posted is the first phase approximating lighting conditions and general model info. Next I will try to build the right textures and model the sink, shower and other components using Cheetah and SketchUp. This may take a couple weeks as my schedule allows.

I do this to learn more about Cheetah's capabilities and show others it's potential. I think there is a school of thought that buying a high-powered and expensive rendering software is the only way to get high-end results.

If anyone has any input or advice I'd be glad to hear what you have to say. Could be a fun project.[/url]

[TrevorM-L]

If any one can do it with C3d... you can.... :mrgreen:

and it looks like you are definitely getting somewhere with it as the basic modelling and rendering that you have done so far is pretty damn good if you ask me...

this is sure to be one heck of a good thread to follow... for if you also post your tips and tricks as you usually do Jake we will all be C3d pros in no time at all... well maybe not for a week or so when you get it done...

Back to work Jake... we are all wating here... :lol: :lol: :lol: :lol: :lol:

cheers and all the best with it...
Trevor

[lavardera]

I do this to learn more about Cheetah's capabilities and show others it's potential. I think there is a school of thought that buying a high-powered and expensive rendering software is the only way to get high-end results.

Cheetah is a high-powered rendering program.

[Jake]

I do this to learn more about Cheetah's capabilities and show others it's potential. I think there is a school of thought that buying a high-powered and expensive rendering software is the only way to get high-end results.

Cheetah is a high-powered rendering program.

Oh, you know what I meant Greg! :wink:

[Francois]

Jake,
With what I have seen from you on various forums, there is no question at all that you can get there with Cheetah3D. The beginning is already impressive !!

[Jake]

Well, at the risk of sounding like the Mutual Admiration Society I think I can safely give credit to Martin for providing an easy to use, yet powerful 3d program.

I need help making a realistic towel that hangs from a bar! I doesn't hang that well. What would be the best technique Martin-or anyone.

[Francois]

Jake,
What about creating one side of it like a topo in SU and then copy/flip horizontaly for the other side ?

[lavardera]

Jake - how about using an image of a hanging towel and making a greatly exagerated bump map of the image?

Ahh - I just looked at it again, its just a little wavy and not very wrinkled.

[Jake]

I''l use a bump to texture it, but there is a bit of a wave. Making that look natural is tough for me.

I was trying to use the relief object. I made a towel in sketchUp and brought it into Cheetah, added subdivisions and tugging at it with the magnet, but it wasn't pulling both side of the towel. It was adding thickness.

I'm not to good at the organic stuff.

[Francois]

Here is maybe another way. Just a quick test to show feasability. It seems to me that with a little more tweaking of the displacement and subdiv parameters you could get there. I think also that a larger gradient should also be used for the displacement map. The taper is there just in case you want to shrink a little the top part

[Jake]

That was my next step Francois, but I have been using the relief object to do the same thing. I'll airbrush a b&w image in Photoshop and drop onto a relief object. That's how I've been doing my contours.

Your way to make the towell is the way I need to learn. I could really use some training when it comes to modeling-especially the soft, organic shapes.

What process would you recommend get the towell to drape over the towell rod in a realistic fashion. How do you get the towell to bend around the bar?

[Jake]

Heres a shot with a few accessories added. I still need to work on the sink at the corners, but it's getting close. There isn't enough shadow under the radiator by the window, but I think I can fix that by moving my fill light forward a bit.

I'm going to get that towel done today using Francois's suggestions.

[lavardera]

Really shaping up Jake.

That's exactly what I was thinking for the towel - if you airbrush the wrinkles it might look better and more flowing. I would do this - create a rectangular image with a gradient from light to dark to light, and then use a deformation tool to neck down the middle or edges so you get a much more broad flowing wave to the bump map.

I'm guessing the wall texture is just a place holder? The bump map is a bit soft and the texture image looks more like stucco where it should be like cast concrete with fairly sharp voids for the bump map.

[Jake]

Really shaping up Jake.

That's exactly what I was thinking for the towel - if you airbrush the wrinkles it might look better and more flowing. I would do this - create a rectangular image with a gradient from light to dark to light, and then use a deformation tool to neck down the middle or edges so you get a much more broad flowing wave to the bump map.

I'm guessing the wall texture is just a place holder? The bump map is a bit soft and the texture image looks more like stucco where it should be like cast concrete with fairly sharp voids for the bump map.

Yea, I'm proceeding with the towel as outlined by you and Francois.

I still need to do the final wall texture. The one I'm using now is a little soft and plastery. That could be a good lunchtime project. I agrre it needs to be sharper.

[lavardera]

I combined these two images to get a blended gradient. I did it very quickly and not very carefully - it needs to be pure white at the top to meet the bar - but you see where I'm going with it. This is a way to make a smooth wavy gradient for a bump map.

[Jake]

I'm sure that would work Greg. Thanks. Looks like Op Art.

My question though is how would you make the actual towel. Would it have a thickness and how would the gradient affect an object with thickness? I think it would only affect the outside surface of the towel.

This seem like a dumb question doesn't it.

[TrevorM-L]

Jake

it is really starting to look good... obviously lots of little tweaking to go on still but definietly coming together as a challenger to the V-Ray render...

Out of curiosity... at this point are you doing much work in SU and exporting it out each time to test it or are you using the editing within C3d to continue with what was started in SU...

How many lights are you putting into the scene to develop all of the intricacies of light play that would be occuring in such a real scene...?

cheers... you are doing a great job... and teaching us all a lot of the technicalities and possibilities of C3d... our great little but not so little app

Trevor

[TrevorM-L]

I'm sure that would work Greg. Thanks. Looks like Op Art.

My question though is how would you make the actual towel. Would it have a thickness and how would the gradient affect an object with thickness? I think it would only affect the outside surface of the towel.

This seem like a dumb question doesn't it.

The towel would definitely have a thickness... maybe actually pull a thickness into it... put your texture onto it... then use a bump map to assist in the modelling of the folds... may have to actually take a photo of a towel on a rail to get the idea of how it sits on the top as it wraps over the rail and how that affects the fall of the towel...

cheers
Trevor

[Jake]

Thanks Trevor,

Here's a wireframe showing the lights. Just two lights counting the HDRI for the sky image-an area light from the window and a point light just inside the window with shadows off to bounce light around the room. I'll be glad to send the file to anyone who wants them. Might be a little big, but if you can handle it, I can send it!

Vray, Maxwell, etc. are great renderers, but I think they are needlessly complex. The object of this exercise is get comparable results with Cheetah. There is no question in my mind that it can be done, but I'm still pretty new to this type of thing.

And with the improvements that Martin is planning, it is only going to get better and easier to get great results.

I'm off to Gary, Indiana to photograph a building. For those of you familiar with Gary, wish me luck!

[TrevorM-L]

I'm off to Gary, Indiana to photograph a building. For those of you familiar with Gary, wish me luck!

I work with a Gary... he can be a bit thick at times but not quite like a building though... :lol: :lol: :lol:

I have no idea about Gary, Indiana... have fun tho'

You can send me the zipped file at some stage... no hurry tho' as you are obviously somewhat busy with this... work... and family...

thanx
trevor

[Francois]

My question though is how would you make the actual towel. Would it have a thickness and how would the gradient affect an object with thickness? I think it would only affect the outside surface of the towel.

Jake,
If the goal is only to duplicate as closely as possible the V-Ray render, then I don't think there is a need to put a thickness on the towell since this will complicate the mesh. I had a close look at the original image and I don't think there is one there. To me, this just looks like 2 single planes attached together on 1 side slightly above the bar.
Beside this, the plane on the back, not being visible, is very likely a flat plane.
Therefore I would just subdiv/displace the front plane and later on add a second plane for the back (+move it slightly to make it visible on the lower right).

Of course, if you want a real towell so several renders can be made from various positions, then it's a different story.

[Francois]

Jake,
Looking at your latest image and more specifically to the towell, I think you need to change the subdivision type from Catmull to linear. I can see that your mesh is not regular. By going to linear I think you will get a more "linear" displacement and the map will be more evenly distributed. At least this is what I have experienced in the quick test I did. But, I'm not an expert ... just trying to help... :wink:

[Jake]

Well, back from lovely Gary, Indiana. The attached image is the sign slapped on the front door of the place I had to photograph. I think it speaks for itself.

Francois,

I'm not trying to duplicate the image exactly. Im trying to get a rendering that is comparable. I need to duplicate stuff in the image though to get an apples to apples comparison of how the light and surfaces react, but it doesn't need to be a carbon copy.

I really would like learn how to model a darn towel correctly.

ps. you are correct about the subdivision. I should toss that rag right into the hamper.

[Jake]

I rendered a shot of the sink and faucet while I was gone. Think it looking pretty good except for the razor sharp edges inside the sink. When I modeled it I expect to see those inside edges. Got lazy.

I need to apply a little smoothing to the sink front. It still looks a little facetted. And of course the wall texure and bump map are pretty bad.

[Jake]

Same image with a touch of Photoshop and some fake depth of field.

[TrevorM-L]

Jeez Jake.... go easy on us will you... you are setting the bar toooo high... especially for a total novice like me... :lol: :lol: :lol: :lol:

those basin shots are great even as they are... but of course you could do a little better tho' if you really tried a bit more... :lol: :lol:

what were your settings for the taps and the mirror... they have both come out looking like photos... very good work indeed...

cheers
Trevor

ps... I can see what you mean now about Gary... oh well... takes all types to make this world I guess...

[Jake]

Trevor,

I'll post the file tomorrow or at I'll the least post the settings. That model is at work. The settings really aren't anything special. Reflection at about .7 and specular set to metal at 100. Use a medium grey for the color. Oversampling pass 1 at 8 or 16. Oversampling pass 2 at 0.

Jake

[TrevorM-L]

Use a medium grey for the color.

ah huh... that is where I must be going wrong with my chrome... I have it set far too light...

for the oversampling... I thought one was to set the 2nd run at the high level and the 1st at the low level... thought I saw it somewhere in the documentation... what is the reason behind this... as I am still not sure what it is fact doing...

cheers
Trevor

[Jake]

It's a cache thing. Maybe Martin can explain it better.

[Martin]

Hi,
you normally have to set "oversampling pass 1" to the lower value. In most cases a value between 1 and 4 should be enough. Nevertheless here is a small description to see how the oversampling algorithm works.

1. In the first pass for every pixel "oversampling pass 1" samples will be taken.

2. The next pass is a edge detection filter which depends on the "tolerance". It detects pixels who have neighboring pixels with different color. The lower the tolerance the more pixels will be classified as edges. Pixels which have beed detected as edges will be highlighted in red.

3. In the last pass those pixels which have been detected as edges (the red ones) will be refined. Therefore an extra "oversampling pass 2" samples will be calculated for that pixel. Since pixels which haven't been detected as edges won't be refined you save a lot of rendering time.


For many cases it is therefore enough if you set "oversampling pass 1" to 1 because the edge detection filter already find the edges. There are only some cases where you need a higher setting for "oversampling pass 1".

I hope that helps.

By,
Martin

[Martin]

I really would like learn how to model a darn towel correctly.


There are probably a hundret ways to model a towel. This is how I would make it. Each step corresponds to the attached image.

1. Start with a box. Set the size and the sections and make it editable.
2. Remove the three sides of the box you don't need.
3. Select the edges according to image 3 and move them a little bit outward.
4. Do the same with the other edges according to image 4. But this time move them inward.
5. Finally apply 1 or 2 subdivision steps with either the polygon tool or the modifier object.

As I said before this is just one way to do it.

By,
Martin

[Jake]

I guess I'm confused then Martin, When I was working on the Farnsworth Rendering there were some artifacts that I though you said would be eliminated by not using the second pass. I though you explained it as a cache issue.

What would be the best settings? 1-4 for the first pass and as high as you can afford to wait on the second pass?

[lavardera]

Gulp!

Sometimes I wonder if I am ever going to get the hang of this rendering thing..?

You experts - there would be a good market here for selling preset scenes that a goof ball like me could drop a model into, and click render..

[Jake]

Gulp!

Sometimes I wonder if I am ever going to get the hang of this rendering thing..?

You experts - there would be a good market here for selling preset scenes that a goof ball like me could drop a model into, and click render..

Yea, well I wish I were goof-ball architect with your design abilities Greg. I think you'll get the hang of Cheetah3D when you get the time to really look at it closely. Right now you need to concentrate on saving the world from McMansions! :lol:

[Allister]

Yeah, I'm with you Greg, they lost me when the over-sampling started ...

These renderings are amazing Jake, good job. I'm also intrigued about the freeform stuff that you can't do that easily in Su ... ;-)

[Jake]

Yeah, I'm with you Greg, they lost me when the over-sampling started ...

These renderings are amazing Jake, good job. I'm also intrigued about the freeform stuff that you can't do that easily in Su ... ;-)

That intrigues me too. So far most of the free form stuff is already done. I can soften and edge or add a subdivision and distrot shapes to make them a bit more natural, but I haven't done a lot of freeform modeling in Cheetah except for site contours.

That are folks in the gallery that do quite well in that department. I'm still too SketchUp dependent. Is there a SA (sketchupper's anonymous) I could join? We could meet as a group each month and learn to draw organically!

Actually Allister. If you plan ahead as to how you group your SketchUp models. You can apply different subdivisions to say, seat cushions to soften the edges. You just need to think ahead about how you group your models or materials.

I have started to play with Martin's modeling tools and they really work well. Maybe I'll tackle a beanbag chair first!

[Atombender]

I can't wait to see the final image!
I love these "tutorials" 8)