Carrara Challenge 3: PARADISE LOST AND FOUND. Work In Progress (WIP) thread.

Varsel

Well I'm curious to if I am going to pull this off......
I have a vision of a picture, but managing to materializing it out isn't always easy.

Texturing the ArchiTools models are actually very easy. They are being, as Holly so beautifully put it, “automagically” created.

But : there is a trap.
As you see in the picture. The direction you are drawing the walls matter. One has to draw them in the same direction all the time.
The two buildings are both two rooms joined together, but the bottom one is one room drawn from left to right, and the other from right to left.
The top building, both rooms are drawn from the left to right.

I believe it has to do with that all rooms have a inside and a outside, and somehow the direction one are drawing them decides wich one is what...?

And UV mapping in the window and door cutout, doesn't work to well. But that doesn't matter anyway. There is going to be something in there anyway.

Szark

I so wish I could join in but with this machine I can't do much in Carrara at all these days. Well not the stuff I like to do that is. ;)

I will say though it sounds a cool challenge Antara and I hope it is successful for eveyone's sake and Carrara's.

pnewhook

Well here's my WIP entry to the contest. Basically it will be a paradise LOST.

Stock object is an M3 skeleton, modified.
Obviously there are just a bunch of scribbles for placeholders as I explore what I'm going to do. This will be my first attempt at something that doesn't have a robot of some sort in it

Sockratease

Antara said:

Sockratease said:

Wow, more than a few questions / comments on those couple NPR renders!

Thanks to those who liked them.

This question seems to cover all the others too, so I'll answer it, and if I missed something, feel free to ask for more info.

These were both single layer and single pass renders, no postwork or layers were needed.

I was frustrated with the way NPR handles things like fire, transparencies/alpha, and effects - so tried to get around them.

Couldn't make anything work for alpha /transparency or effects - but found a silly work-around for fire.

That is not Carrara fire. Joe Mamma would like this technique because I took a picture of fire (HA! - Using photos to make NonPhotoReal stuff!) (The irony is not lost on me) and brought it into PhotoStudio 5.

Then cut out the background so it was just fire and nothing else. Save it out as a png, jpg with black background, and greyscale.
I used the greyscale as a terrain height map and got an ugly terrain whose contours kind of matched the fire. A little.
Then converted the terrain to a vertex object and trimmed off the flat square ground plane surrounding the raised bits.
Then gave it a simple "Box" UV map and used the png for color.

It rendered quite nicely for NPR fire considering Carrara's native fire wont render at all!

The jpg wound up being useless - it was just in case I needed it for other reasons (the png had alpha and thought it could be a mask, but trimming the model was easier than expected, so I fed the jpg to my cats - they love eating computer files!) (how else could I lose so many?).

The fire looking so different in both renders, despite being the same obj file, comes from it's angle. It had stupid high peaks and would look silly in a realistic render - but that works to advantage in NPR. The first one is facing the camera and the second one is angled slightly away. The simple techniques often are the best.

As for the "different brushes" look - I was experimenting to see what NPR responds to, and does not respond to. Turns out the bump channel is a major player with NPR. I set some objects to very smooth and others outrageously high. The big yellow bird has the bump set as high as possible, and I even used a multiply operator on the bump map (I think it was the 10,000% cranked all the way up) and it got a nice organic look.

As usual, the background was a fractal, and it rendered out nicely as an abstract splash of colors.

I think that covers it.

Gotta dig around for more of these type renders. I know I have quite a few NPR's in the pile which the cats never got to...

Your fire solution is pure genius! On so many levels!

I really like the idea of creating the mesh out of a BW photo or map. I can see that being handy for a much wider range of applications. Totally brilliant idea!!! I'm not good at modelling at all, so this is just the type of solution I might find actually doable... And then you can use boolean functions to create a single symmetrical object out of the 2 halfs created from such terrains.... I'll have to try it some time... And I can use stamp-type brushes for the original terrain maps... So much potential!

And thank you so much for the bump tip. I was getting very tiny brush size and smooth looking results on my items with bump and kept thinking it's due to polygon sampling, so I kept decimating them, with little to no difference.

Also, are these images done with just diffuse layers? Have you played with good ways of combining different layers?

Modeling from images is an old favorite trick of mine. Works great with vectors and the spline modeler too.

I have quite a few - more to excavate!

And yes, just diffuse layers. Nothing fancy.

If I want to layer, I usually make two identical sized renders and mask out bits of one to make a coherent whole (like if I want a different brush style for one object, I'd layer it on top and make a totally black mask to hide the whole layer, then "paint in" my object by brushing white to reveal it with a soft edge brush so the blend is less obvious).

I forgot to answer the question about light through transparency! I usually cheat and just use the image I want where the transparency would be - like pasting the eyeball image over the transparent layers in an eye. If it's a window or other aperture, just delete the glass - nobody will ever know!

ManStan

I'm running in to issues with my next entry, soft cloth really bogs this comp way down. Once again the theme is lost and found.
So far it is the M3 pirates clothes, the old P4 skeleton, and the treasure chest from DAZ. The rest will include several DAZ figures, but the island and flora will all be carrara. I'm going to try to modify all the clothes with draping. But that may be beyond this comp. I may have to do each piece of clothing individually, save out as a .obj then import refit and retexture. Fun stuff.

edit to add. Working fine now, kept getting memory allocation warnings last night. I dropped my undos from 16 to 4. I think that was the main issue for this old comp.

evilproducer

ManStan said:

I'm running in to issues with my next entry, soft cloth really bogs this comp way down. Once again the theme is lost and found.
So far it is the M3 pirates clothes, the old P4 skeleton, and the treasure chest from DAZ. The rest will include several DAZ figures, but the island and flora will all be carrara. I'm going to try to modify all the clothes with draping. But that may be beyond this comp. I may have to do each piece of clothing individually, save out as a .obj then import refit and retexture. Fun stuff.

edit to add. Working fine now, kept getting memory allocation warnings last night. I dropped my undos from 16 to 4. I think that was the main issue for this old comp.

Hey Stan,

That's a great idea about dropping the levels of undo when running into memory issues. I had never thought about before. I will have to make a note of that so that I can try it the next time I get a memory error. Thankfully, it happens very rarely with me.

ManStan

1 gig machine, happens a lot, lol. Well not since I reset the undo.

This took most of the afternoon. Needs retexturing but the draping came out nice. Does that count as moding content? :coolsmirk:

evilproducer

ManStan said:

1 gig machine, happens a lot, lol. Well not since I reset the undo.

This took most of the afternoon. Needs retexturing but the draping came out nice. Does that count as moding content? :coolsmirk:

I would certainly hope it would count. You're using the bullet physics engine in Carrara to do it. ;-)

Headwax

just a quickie :0
be back tomorrow and admire everyone's work

Antara

OK, I have to stop playing with NPR. This thing is freaking addictive.

But before I move on to other things, here is my best effort at it so far.
I think the best way to do it is to combine layers from NPR and the regular Photo-realistic multi-passes. Which is what I did for the combined image. (I attach the layers showing their modes, modifications and opacity (in %) to show what the final look is made of.)

I like using ornate, but light gery brushes for this because I think they give the most freedom for post. Also I found that i like brushes, which are small (compared to the white space around them within the brush square), and I used light grey to slightly darker grey for the brush tone, so that the overall layer pass would not be too bald. I like multiplying layers more than reducing opacity on them. Also when such a layer is blurred slightly and then multiplied, it creates a more painterly look.

The regular multi-pass was rendered in much smaller size, since for this type of image precision is not at all necessary, so it was nice and fast. Low anti-aliasing settings throughout. I don't see much difference there, maybe because of not using outline. I resized the photo-realistic render in post to match my NPR layers size.

I also used Photoshop watercolor filter on most photo-realistic layers. Just because it looked more fun that way :) And then I blurred them to reduce the hard edge effects.

And hopefully now I will be able to tear myself away from NPR to start thinking about my entry....

Now to other topics:

zgock said:

thanks.

YAToon Settings of WIP1&2 are almost Same.
Based on My YAToon Tutorial Shader and set Shadow Softness to 50%ish.
and add a bit Specular Softness(you should adjust this value with your lighting)
and set Shadow Shader to Simple Red Color Shader.

and added SSS with dark brown.

In WIP2, I add Skylight 60%ish and ambient to 0%
(WIP1 was ambient 10%)

Posted NPR Render was rendered with C6 about 4 years ago. so I forget settings sorry.
but I can Render new NPR Image in nearly setting.
please gimme a sec.

Next Step is Adding Background to WIP2. Maybe I can show you settings you need.

Thank you so much for the screenshots! YAToon is next on my list of things I'm trying, though I just had time to install it and download the tutorial files today, so it will have to wait until Monday or whenever I can carve out some time for it.

I would never have thought of using SSS with the toon shader, neither would it occur to me to use red color for shadow... Therefore, seeing your settings here will be a huge help, thank you!

I can see you are using some very advanced settings here, the sky light and the SSS are usually pretty render-time expensive. And the 0.5 object accuracy and also slow things down. You also mentioned that you render images at different size than your intended final result...

How big are the images you normally render and how long does it usually take?

Also, did somebody already help you with the tutorial? I started following it and it reads quite naturally.

Varsel said:

Well I'm curious to if I am going to pull this off......
I have a vision of a picture, but managing to materializing it out isn't always easy.

Texturing the ArchiTools models are actually very easy. They are being, as Holly so beautifully put it, “automagically” created.

But : there is a trap.
As you see in the picture. The direction you are drawing the walls matter. One has to draw them in the same direction all the time.
The two buildings are both two rooms joined together, but the bottom one is one room drawn from left to right, and the other from right to left.
The top building, both rooms are drawn from the left to right.

I believe it has to do with that all rooms have a inside and a outside, and somehow the direction one are drawing them decides wich one is what...?

And UV mapping in the window and door cutout, doesn't work to well. But that doesn't matter anyway. There is going to be something in there anyway.

Thank you for the explanation. Now I am even more impressed with the plug-in. It's one of the few Carrara plug-ins I don't yet have, and the reported UV-mapping issues was one of the reasons I didn't buy it. Now it's time to reconsider :).

I am nowhere near as good at modelling as you are, though, so I'll have to do more research first.

I am also curious about how poly-count-intensive the resulting models are. With your scene, I'm suspecting that you'd be introducing a bunch of figures (even if they are low-res humans) and plants, so I expect polygon count will at some point become an issue...

I like your progress very much and I am very curious what your idea is going to be.

pnewhook said:

Well here's my WIP entry to the contest. Basically it will be a paradise LOST.

Stock object is an M3 skeleton, modified.
Obviously there are just a bunch of scribbles for placeholders as I explore what I'm going to do. This will be my first attempt at something that doesn't have a robot of some sort in it

This looks very promising. Great concept, and I like the skeleton partially under water idea. That should call for some interesting shaders :)

Modeling from images is an old favorite trick of mine. Works great with vectors and the spline modeler too.

I have quite a few - more to excavate!

And yes, just diffuse layers. Nothing fancy.

If I want to layer, I usually make two identical sized renders and mask out bits of one to make a coherent whole (like if I want a different brush style for one object, I'd layer it on top and make a totally black mask to hide the whole layer, then "paint in" my object by brushing white to reveal it with a soft edge brush so the blend is less obvious).

I forgot to answer the question about light through transparency! I usually cheat and just use the image I want where the transparency would be - like pasting the eyeball image over the transparent layers in an eye. If it's a window or other aperture, just delete the glass - nobody will ever know!

Ok, this is very interesting: Please elaborate about vectors and the spline modeller. How does that work?

BTW, I haven't tested it and I am trying to stop my NPR obsession, but I have a question: if you have your brush orientation set at 100% or some other fairly high value, and you render 2 renders with all the settings identical, sill you still get slightly different results, because all strokes will have slightly different orientation due to randomization? or is it fake randomization that gets repeated on same settings? What if you slightly change the randomization percentage? will that shuffle the brush stroke orientation?

Also, I am confused about light through transparency. I had a black-diffuse-color transparent glass bottle in my image originally, with transparent liquid. But I changed the diffuse colors to blue for the bottle and to red for liquid for the purposes of NPR. Since NPR does not do transparency, I expected my liquid not to show up at all. And it doesn't show color-wise. But it does show me the level of the liquid in the bottle, as you can see in the renders. So that means that it does handle transparency to some extent. I just can't figure out how... Do you have any ideas about it?

1 gig machine, happens a lot, lol. Well not since I reset the undo.

This took most of the afternoon. Needs retexturing but the draping came out nice. Does that count as moding content? :coolsmirk:

WOW!!! That's great use of dynamics! And no kidding it counts! If the shirt and pants have been dynamically draped in Carrara, then you already have 2 items towards you 3-item count.

Great job on the draping! Are you using a low-poly proxy figure for the skeleton? I thought it's be near-impossible to drape on something so small-detailed and high-poly. Also are you draping the mesh and then smoothing it for the render, or are you cloth items high-poly too? they look really smoothly draped.

I look forward to seeing where this image goes from here. So far it looks fun :).

just a quickie :0
be back tomorrow and admire everyone's work

That was an unexpected twist. :) Now it's a whole new story I want to hear...

P.S. It's a pity I cannot insert images directly into the body of the post (or is there a way to do it?) because now, all the images which are relevant to the very beginning of this extremely long posting are going to show up at the very end... :long:

evilproducer

Thanks for posting that Antara! Tons of good information there about using the NPR, render passes and compositing! I think pretty soon you'll be able to hold your own against Dart for post length as well! ;-)

I've added hair-as-grass to my highway. I'll have to darken the colors used in the hair shader as the colors in the grass are rather electric, to say the least.

ManStan

It's the old P4 skeleton. Still works :lol:
I'm working on the beach/island made in carrara. I'm trying to make a hirez beach texture in Genetica, but hirez and one gig just wont work; the advantage is a real normal map ;-) .

Next up is to err, umm, distress the pirates essential boat. :coolgrin:

Diomede

Some great goings on while I've been away from the web. Manstan, I like the use of physics. Wish I had a better handle on that.

Antara, thanks for the great NPR posts. I have a question about creating and saving custom NPR brushes (how?), but I put it in a new thread because it seems like a more general issue than just this challenge.

Antara

evilproducer said:

Thanks for posting that Antara! Tons of good information there about using the NPR, render passes and compositing! I think pretty soon you'll be able to hold your own against Dart for post length as well! ;-)

I've added hair-as-grass to my highway. I'll have to darken the colors used in the hair shader as the colors in the grass are rather electric, to say the least.

Depending on what style you are going for in your final image, electric grass might be just the thing :). But seriously, consider just darkening the roots in the hair shader. Also, this way, if you have some noise in the length channel of the shader, the darker roots will show up where neighbouring grass is shorter creating more sense of grass depth and diversity. And I really love how the grass already looks at the outer sides of the road - very natural!

And I am getting more and more curious about your idea.... :)

ManStan said:

It's the old P4 skeleton. Still works :lol:
I'm working on the beach/island made in carrara. I'm trying to make a hirez beach texture in Genetica, but hirez and one gig just wont work; the advantage is a real normal map ;-) .

Next up is to err, umm, distress the pirates essential boat. :coolgrin:

Is the old P4 skeleton low-poly? I don't think I've worked with it. But you are right, it works really well and looks good.

Well, add in that terrain and you'll already pass the 3-item criteria. And if you create debris using Carrara modifiers, that would also count as your modified items, even if the original object is content.

Good luck with terrains and boats! :)

Some great goings on while I've been away from the web. Manstan, I like the use of physics. Wish I had a better handle on that.

Antara, thanks for the great NPR posts. I have a question about creating and saving custom NPR brushes (how?), but I put it in a new thread because it seems like a more general issue than just this challenge.

I have answered to the best of my knowledge, and since this thread is already full of NPR discussions, here is the link to your thread, so people wouldn't need to go searching: http://www.daz3d.com/forums/discussion/27068/

evilproducer

Antara said:

evilproducer said:

Thanks for posting that Antara! Tons of good information there about using the NPR, render passes and compositing! I think pretty soon you'll be able to hold your own against Dart for post length as well! ;-)

I've added hair-as-grass to my highway. I'll have to darken the colors used in the hair shader as the colors in the grass are rather electric, to say the least.

Depending on what style you are going for in your final image, electric grass might be just the thing :). But seriously, consider just darkening the roots in the hair shader. Also, this way, if you have some noise in the length channel of the shader, the darker roots will show up where neighbouring grass is shorter creating more sense of grass depth and diversity. And I really love how the grass already looks at the outer sides of the road - very natural!

And I am getting more and more curious about your idea.... :)

That's a good idea about the shader. I've been experimenting with it, and have desaturated the colors and darkened them in both the root and tip channels. I'll try darkening the roots even more to see what effect that will have. I've also found lowering the highlight and shininess from the default 20% has helped as well.

Headwax

boy blink and you miss stuff around here

firstly Antara - you are doing a wonderous job thank you, I'm just about to read your stuff on the NPR and take some notes :)

Varsel wrote

But : there is a trap.
As you see in the picture. The direction you are drawing the walls matter. One has to draw them in the same direction all the time.
The two buildings are both two rooms joined together, but the bottom one is one room drawn from left to right, and the other from right to left.
The top building, both rooms are drawn from the left to right.

I believe it has to do with that all rooms have a inside and a outside, and somehow the direction one are drawing them decides wich one is what…?

And UV mapping in the window and door cutout, doesn’t work to well. But that doesn’t matter anyway. There is going to be something in there anyway.

thanks for that discovery :)
I played with architools initially but havn't had a chance to delve into it, but I noticed I screwed up the uvees somehow and maybe that's why!
so good spotting. :)

Headwax

Antara, thanks for the NPR stuff very interesting. Your layers look like mine except I never name mine :) which tells you about how confused I can get. OH and I never use the adjustment layers which is silly of me :)

ManStan, I love your concept on this one :)
And Evil, the road is looking so good . I wouldn't stand on it for fear of becoming road kill.
pnewhook: great how you are treating this in the traditonal sense of sketch then render.
Socratease : moo :) Thanks for the hints on the transparency stuff.

From the great results so far I think I am subtly persuaded to try out the NPR again.

Headwax

well I;ve decided that I think by doodling. The concept of this is that the little girl is losing her childhood innocence - (not to the man) signifed by the way she is looking at the moon, and her abandoned Teddy Bear in the background. The man on the other hand is rediscovering his childhood. The dog is a blind dog that has lost it's master - who is nowhere to be found :) The whole setup is like a stage play

Yeah, I know it's s stretch.
Amongst other things...
I used PhilW's wonderful Linton Hall, from English Village. The curtains and newspaper are from Age of Armours' equally wonderous Studio Paris. Both come with a surfeit of props and I would highly recommend them . Both Figures are K4. The K4 on the right is the neanderthal morph and texture. The clothes are an m4 set I refitted .

So far I made the sign and if I go with this concept I'll make a few more things and say exactly what is what.
I include the before post work image as well.

cheerios :) thanks for looking

evilproducer

head wax said:

Antara, thanks for the NPR stuff very interesting. Your layers look like mine except I never name mine :) which tells you about how confused I can get. OH and I never use the adjustment layers which is silly of me :)

ManStan, I love your concept on this one :)
And Evil, the road is looking so good . I wouldn't stand on it for fear of becoming road kill.
pnewhook: great how you are treating this in the traditonal sense of sketch then render.
Socratease : moo :) Thanks for the hints on the transparency stuff.

From the great results so far I think I am subtly persuaded to try out the NPR again.

I agree! The road will become much busier soon!

evilproducer

Looking good Head Wax. I think you inadvertantly gave me an idea for a second entry if I have the time!

Varsel

Well here goes. I'm trying to create the landscape.
Also made a new gate for the city. Just a bit of Asian flare to it. After all Shangri-La is supposed to be the Himalayas somewhere.

I'm not happy with the landscape. It lacks drama.

Any ideas anyone ?

Philemo_Carrara

Not yet a Wip, but with all those talks about NPR and the challenge theme, I would like to try (at last) something that had been nagging me for some time : Emulating engraving.

So, I'll try a tribute to Gustave Doré work :
Gustave Doré illustration of Milton's Paradise Lost

I'm starting the Proof of Concept now and will post as soon as I get something showable (if ever :-) )

Sockratease

Varsel said:

...I'm not happy with the landscape. It lacks drama.

Any ideas anyone ?

It needs a Giant Cow peeking over the hilltops at the camera.

Sticking it's tongue out.

Stezza

I have an idea for mine but the hardest part is the modelling.. I really suck at it, that's why I buy stuff!!

Anyways I have tried to model a few things for this one, a segway because that's what popped into my mind along with some telegraph poles, wires and a sign post..

Putting these together for a render with my thought pattern wasn't difficult, whether others will see what I do is another question lol

The Norfolk pine is a photo I took with my iPone4S and then stuck it onto a plane.. I told you I really suck at modelling lol

Here is my W I P before I post the final render which I played around with lighting and such and it will have no postwork added.

evilproducer

Varsel said:

Well here goes. I'm trying to create the landscape.
Also made a new gate for the city. Just a bit of Asian flare to it. After all Shangri-La is supposed to be the Himalayas somewhere.

I'm not happy with the landscape. It lacks drama.

Any ideas anyone ?

I think the issue with the landscape lacking drama is that it doesn't have the scale you need. The near and mid-level of the terrain look fine, but it's clear that the white capped mountains are coming out the same mesh. They look way to close. If there is a way to keep the foreground and mid-ground aspects, that would be good, but you need a mountain range with real scale for the background.

evilproducer

Here's the highway for my scene. There's going to be a visual gag with it, so I blurred the highway signs as I don't want to ruin the joke before the final image is posted.

Disclaimer: It will be more than be a chuckle type joke ( I hope) rather than a guffaw type joke. Just so I don't build too much anticipation for it, y'know. ;-)

evilproducer

Sockratease said:

Varsel said:

...I'm not happy with the landscape. It lacks drama.

Any ideas anyone ?

It needs a Giant Cow peeking over the hilltops at the camera.

Sticking it's tongue out.

You probably don't want to know what we did to one of our bovines today.

A hint: His name was Whopper Jr. and he was two yrs old.

Stezza

Title - Crossroads

I modelled the segway, the sign post, the telegraph poles and wire and also photographed the Norfolk Pine.

Content credits - LisaB - Indian Grass, SBRM pelican and flock formations

General process description - I used the wide angle camera, positioning and lighting processes for sunlight, GI and full raytracing for render. No postwork was used.

Interpretation - open to the viewers own imagination... but I reckon if you stay under that post to long trying to decide you will be shat upon ;-)

evilproducer

Nice job Stezza! I like the camera angle you chose.

Sockratease

evilproducer said:

Sockratease said:

Varsel said:

...I'm not happy with the landscape. It lacks drama.

Any ideas anyone ?

It needs a Giant Cow peeking over the hilltops at the camera.

Sticking it's tongue out.

You probably don't want to know what we did to one of our bovines today.

A hint: His name was Whopper Jr. and he was two yrs old.

The Millennium Cow will be Immune to that!

But Cows are meant to inspire Art, and if you consider that Art... fine.