I apologize for the lack of updates, folks! Work has been a little slow on the model, I'm afraid. I started running out of things I could do for Soarin', and I wasn't feeling like working on the main model, and then I was struck by a need to start painting in Photoshop again. So I've been doing a lot of 2D artwork of late, and not much 3D. (If you'd like, you can check out some of my recent paintings here.) I also recorded and posted some piano recordings to my YouTube channel yesterday. If you like Star Trek, Michael Giacchino, and/or piano renditions of film scores, you should check out these 5 Star Trek recordings, starting with the one simply entitled..."Star Trek." The others are longer, mind you.
But through all these side project, fear not - history tells us that I always come back to the Disneyland 3D model before too long. As you may have noticed in the CURRENT FOCUSES box, I've put some work into an interesting little side project, too. The Shipley-Lydecker house was a real mansion in Baltimore that was demolished sometime between 1957 and 1971, a photo of which was the primary inspiration for the exterior of the Haunted Mansion. I've matched the camera angle of that one good photo of the house and built a model of it, using aerial photos to make sure its scale is identical to that of the real model. This will allow me to do interesting comparisons between the house and the Haunted Mansion facade, as well as simply see the house from lots of different angles. I hope to do another post soon with lots of renders.
On a side note...I ordered myself a 3Dconnexion SpaceNavigator the other day! I first tried out one of these little beauties in the "Holodeck" at Google's campus, and I instantly fell in love with how it enables you to navigate through 3D spaces in a fluid, dynamic, flexible way. It's compatible with all the 3D apps I currently use with any frequency (Blender, Google Earth, Bing Maps 3D, SketchUp, NASA World Wind), and I intend to make constant gleeful use of it. Only problem is that it probably won't arrive till the end of the week. =P It'll actually be a bit of a race to see which will arrive first - my SpaceNavigator, or the new Haunted Mansion soundtrack that I'd ordered over DelivEARS a few days earlier?
Monday, September 28, 2009
Wednesday, August 19, 2009
More Soarin'
Well, what the heck. I'm still working on Soarin'. My first order of business after the last post was adding the pairs of columns that support each theater. I then textured the screen with the blue lighting scheme Disney uses during loading and unloading. I had to reproduce that look, since there aren't any photos of the whole screen due to its insane size, and the light is poor in there anyway. So I collected reference photos and then put together this image in Photoshop from scratch:
Pretty close, n'est-ce pas? I UV-mapped this to the domed screens and set up their material so that lights in the model wouldn't affect it. (With this setup, you can also see the image from the back of the screen, which is how the screen actually is in real life, or so I'm told.) I also added the black walls that surround the screen, and rudimentary versions of the walls of part of the queue between the two screens. The floor of the loading areas received some preliminary texturing - a shiny, off-white concrete look - and I also built the lower floor next to and under the screens, which I believe is a shiny black. You can see all these changes in this render:
But that's not all, folks. See, I started getting this hankering to add a rather important part of the attraction: the ride vehicles themselves. Although the US patent office has some pretty interesting schematics of the hang gliders, they're not quite the same as the ones Disney built for the attractions, so I found appropriate photos and more or less traced over them in 3D. Now all the vehicles sit in place as they should. I did cheat a little, though...two of the rows in each theater are supposed to have nine seats instead of ten, but to keep things simpler, I just squeezed those same ten seats to be a little narrower. I'm lazy! =P
The other changes visible in this render are the safety railing along the front of the loading area and a large grid texture applied on top of the concrete-looking floor. Luckily, I was able to reuse a grid image I'd made for some other texture in the main model, so, in the words of a certain famous button, "that was easy!"
What won't be easy, however, is adding all the machinery above the vehicles that lifts them up toward the screen. No idea if I'll even make an attempt at that. However, I would like to set up a basic animation so that the vehicles will levitate up into their appropriate positions. More on that as things progress.
Pretty close, n'est-ce pas? I UV-mapped this to the domed screens and set up their material so that lights in the model wouldn't affect it. (With this setup, you can also see the image from the back of the screen, which is how the screen actually is in real life, or so I'm told.) I also added the black walls that surround the screen, and rudimentary versions of the walls of part of the queue between the two screens. The floor of the loading areas received some preliminary texturing - a shiny, off-white concrete look - and I also built the lower floor next to and under the screens, which I believe is a shiny black. You can see all these changes in this render:
But that's not all, folks. See, I started getting this hankering to add a rather important part of the attraction: the ride vehicles themselves. Although the US patent office has some pretty interesting schematics of the hang gliders, they're not quite the same as the ones Disney built for the attractions, so I found appropriate photos and more or less traced over them in 3D. Now all the vehicles sit in place as they should. I did cheat a little, though...two of the rows in each theater are supposed to have nine seats instead of ten, but to keep things simpler, I just squeezed those same ten seats to be a little narrower. I'm lazy! =P
The other changes visible in this render are the safety railing along the front of the loading area and a large grid texture applied on top of the concrete-looking floor. Luckily, I was able to reuse a grid image I'd made for some other texture in the main model, so, in the words of a certain famous button, "that was easy!"
What won't be easy, however, is adding all the machinery above the vehicles that lifts them up toward the screen. No idea if I'll even make an attempt at that. However, I would like to set up a basic animation so that the vehicles will levitate up into their appropriate positions. More on that as things progress.
Tuesday, August 18, 2009
Inside Soarin' over California
Well, the main model has just gotten so clogged due to ironwork and large images and so forth that I've been taking a bit of a break from it. The good news is that I've done some digital painting instead, which has been fun, and now I've started the interior of Soarin' in a different file, just for kicks. This is especially helpful because I've been slowly developing an alternate concept for the exterior, queue, and loading area.
Now, I'm working from blurry blueprints of the WDW version, so if I start getting too detailed, there are bound to be mistakes, but the interiors of both versions' main show buildings are identical in size and shape, so everything I've got so far should be pretty spot-on. One thing I'm not too confident in, however, is the shape of the screen. The overall scale and curvature are correct, and I THINK the bottom half is totally spot-on, but I'm not positive about the way the edges bend forward at the top. If you've got photos, blueprints, personal observations, or anything else that might help me understand the shape of the screens more completely, your help would be greatly appreciated. =D
Notice how the two theaters are just mirror images of each other. The queue will run between them and then along their outer edges. You can already see the long, rectangular structures that I believe house the projectors.
One thing I've learned already is that the Anaheim version is half underground! I never really thought about the fact that you enter the building at ground level and then continually descend a lot of ramps until you enter the loading area. But seeing the blueprints made it obvious that nearly half the building's height is below ground level. (This is not the case in the Epcot version, however, which is located all aboveground.) In fact, when you first walk into the building, half of each screen is lower than the level where you are.
Anyway, I'm done with Pirates or anything. I just need some time to mull over how I'm gonna coerce my machine into handling the ever-increasing complexity of the model.
Now, I'm working from blurry blueprints of the WDW version, so if I start getting too detailed, there are bound to be mistakes, but the interiors of both versions' main show buildings are identical in size and shape, so everything I've got so far should be pretty spot-on. One thing I'm not too confident in, however, is the shape of the screen. The overall scale and curvature are correct, and I THINK the bottom half is totally spot-on, but I'm not positive about the way the edges bend forward at the top. If you've got photos, blueprints, personal observations, or anything else that might help me understand the shape of the screens more completely, your help would be greatly appreciated. =D
Notice how the two theaters are just mirror images of each other. The queue will run between them and then along their outer edges. You can already see the long, rectangular structures that I believe house the projectors.
One thing I've learned already is that the Anaheim version is half underground! I never really thought about the fact that you enter the building at ground level and then continually descend a lot of ramps until you enter the loading area. But seeing the blueprints made it obvious that nearly half the building's height is below ground level. (This is not the case in the Epcot version, however, which is located all aboveground.) In fact, when you first walk into the building, half of each screen is lower than the level where you are.
Anyway, I'm done with Pirates or anything. I just need some time to mull over how I'm gonna coerce my machine into handling the ever-increasing complexity of the model.
Friday, July 31, 2009
Technical difficulties
Well, I've got some good news, and I've got some bad news. The good news is that the ironwork on the front of the Pirates facade is basically finished - excluding the stuff on the curved stairways, that is. In fact, here's the proof!
The bad news is the reason why you're seeing Blender's non-rendered preview here instead of a fully rendered image. Uh...Blender is crashing when I try to render views of the facade, and it's the ironwork's fault. This is, as you can imagine, problematic. And I'm not sure what I'm going to do about it. Worst case scenario, I hide the ironwork and work on other stuff until I get better hardware someday. But I'd rather not have to hide all this hard work I've done...I had to correct numerous photos for perspective distortion, line them up in the model, and manually trace over them in Blender.
Sigh...we'll see what happens. I've certainly got plenty of other stuff I can work on.
The bad news is the reason why you're seeing Blender's non-rendered preview here instead of a fully rendered image. Uh...Blender is crashing when I try to render views of the facade, and it's the ironwork's fault. This is, as you can imagine, problematic. And I'm not sure what I'm going to do about it. Worst case scenario, I hide the ironwork and work on other stuff until I get better hardware someday. But I'd rather not have to hide all this hard work I've done...I had to correct numerous photos for perspective distortion, line them up in the model, and manually trace over them in Blender.
Sigh...we'll see what happens. I've certainly got plenty of other stuff I can work on.
Wednesday, July 29, 2009
The ironwork saga continues!
Sorry for yet another posting that's high on Blender technicalities and low on Disneyland renders, but I think I've decided what I'm going to do about the ironwork. It's a compromise, of course, but until I'm a wealthy man with a render farm in the back room, it'll probably have to do. I WILL make meshes that follow the actual shape of the ironwork, rather than UV-mapping PNG images onto them with alpha, but I'm not going to give them thickness. That'll reduce the number of polys (and thus the render time), but at least from a distance, the effect will be about the same. The major disadvantage is that if I get up close, it'll probably be apparent that the iron is thinner than finest tissue paper. This is not an issue with fully 3D meshes...or with UV-mapped planes, since I would simulate shading in the image texture. But this will hopefully be the best compromise. I cannot possibly express how nice it is to not have to light the balconies and verandas separately.
Now I just need to get up the gumption to start working on the hardcore second-floor ironwork! Gulp...if anyone's got some especially good photos of that area, I'd be much obliged, though I'll hopefully be just fine with what I've already got.
Now I just need to get up the gumption to start working on the hardcore second-floor ironwork! Gulp...if anyone's got some especially good photos of that area, I'd be much obliged, though I'll hopefully be just fine with what I've already got.
Tuesday, July 28, 2009
Ironwork test results =/
Well...I did my little test of render times for the two different ways of creating the ironwork. I must say, I'm disappointed with the results. To recap: I can create ironwork either by applying an image with transparent areas to a flat plane or by creating the actual geometry of the ironwork in 3D. Both ways are labor-intensive for me; I was just wondering which way would render faster. If this test is any indication, the first route is a lot faster. I did two renders of the same spot with the two different techniques. The UV-mapped image technique clocked in at 7:59...
...and the real 3D geometry technique clocked in at a whopping 22:23. =/
This is a disappointment because creating the actual geometry is a rather more elegant solution. For one thing, it lets me actually see the ironwork while I'm navigating through the model; it doesn't just magically appear at render time. Furthermore, it just plain looks better. And perhaps most importantly of all, real geometry allows my ambient occlusion lighting to bounce through the holes and allow us to see the shadowy areas of the balconies. If I create a flat plane with a transparent image texture, ambient occlusion treats it as an opaque wall, and the balcony turns black. Then I have to carefully light the balconies separately, which in turn increases the render time...something I did not do for this test.
I'm using that last factor as a huge excuse because...well, I don't want to UV-map these puppies anymore. I want real geometry. I did the UV-mapping for the HM ironwork, and the results were merely satisfactory after a buttload of tedious work. Render times were still high, and I didn't reap any of the benefits of doing true detailed meshes. So I'm gonna be adventurous and make all the POTC ironwork the more authentic way...and render times can walk the plank, for all I care! =D
By the way...sorry for the weird way the lamp is cut off in the first image. I forgot to activate ray transparency, which also would have boosted render times a little, I think. Another excuse to do real geometry! ;)
Lastly, to my less CGI-oriented readers, I apologize for these more technical posts. I'm sure some of you just want the Disneyland stuff, not the nitty-gritty details of how I do it, but I'll continue trying to deliver both in proper proportions. =)
...and the real 3D geometry technique clocked in at a whopping 22:23. =/
This is a disappointment because creating the actual geometry is a rather more elegant solution. For one thing, it lets me actually see the ironwork while I'm navigating through the model; it doesn't just magically appear at render time. Furthermore, it just plain looks better. And perhaps most importantly of all, real geometry allows my ambient occlusion lighting to bounce through the holes and allow us to see the shadowy areas of the balconies. If I create a flat plane with a transparent image texture, ambient occlusion treats it as an opaque wall, and the balcony turns black. Then I have to carefully light the balconies separately, which in turn increases the render time...something I did not do for this test.
I'm using that last factor as a huge excuse because...well, I don't want to UV-map these puppies anymore. I want real geometry. I did the UV-mapping for the HM ironwork, and the results were merely satisfactory after a buttload of tedious work. Render times were still high, and I didn't reap any of the benefits of doing true detailed meshes. So I'm gonna be adventurous and make all the POTC ironwork the more authentic way...and render times can walk the plank, for all I care! =D
By the way...sorry for the weird way the lamp is cut off in the first image. I forgot to activate ray transparency, which also would have boosted render times a little, I think. Another excuse to do real geometry! ;)
Lastly, to my less CGI-oriented readers, I apologize for these more technical posts. I'm sure some of you just want the Disneyland stuff, not the nitty-gritty details of how I do it, but I'll continue trying to deliver both in proper proportions. =)
Sunday, July 26, 2009
The tribe has spoken!
Despite bitterly polarized opinions on the matter (wink wink), I've made my decision about what will be occupying the space above the entrance to Pirates. =D I consider the new creation in this small render somewhat of a down payment:
Thank you very much to everyone who contributed their surprisingly unanimous input! You made me feel a lot more confident as I made my decision, and I'm glad y'all apparently want the same choice I was leaning toward.
Actually, while we're on the topic of signs, I may as well explain how I go about making them. For some simple signs (an EXIT sign in the Indy queue comes to mind), all I really do is find a good photo of it and then slap that bit of the photo onto an object of the right size and shape. But for stuff like the entrance signs for the Dream Suite and Pirates, I prefer a somewhat more robust solution. The thing about photos is...they're accurate, but their resolution is limited. And in order for the image to appear sharp when it's rendered by Blender, it needs to have fairly high resolution. (Besides, you can get yucky stuff like noise and perspective distortion and objects in the way when you're dealing with photos.)
So my labor-intensive but oh-so-definitely-worth-it solution is to replicate the sign in Photoshop using appropriate fonts, vector techniques, and layer styles. I'm basically tracing over one or more perspective-corrected reference photos using the pen tool at first...then it's a matter of getting the colors and materials to look right. It takes a while - particularly tracing - but the result is a very sharp, clean image that can then be UV-mapped to the sign object in Blender. (Just doing a non-UV texture works too, since the sign object is typically flat and only one side of it shows, but that doesn't allow me to see the image previewed on the object as I navigate through the scene.)
As a couple of extra goodies to illustrate what I mean, here are my texture images for Pirates and the Dream Suite. In the case of these two signs, I've built the borders as separate, truly 3D objects that just take flat color materials; the UV-mapped sign object sits nestled within the frame, so you don't see that frame in the image. And notice how the corners don't show up on the final sign objects, since they have rounded shapes, so it doesn't matter what goes in the corners of the image. Also note how I have to simulate elements that would be truly three-dimensional in real life, such as raised lettering. I've also added a reflection to the Dream Suite image so that I don't have to increase render times by doing a truly reflective material. Finally, I want to point out that although the Dream Suite sign is way physically smaller than the Pirates sign, I've made the Dream Suite image significantly bigger so that the small Disneyland lettering is still sharp. If I wanted to get really fancy, I could put that in a smaller, higher-res image and let the rest of the image have a lower resolution, but I guess I'm just not that fancy.
Also, just so ya know, I've now added the curved steps on either side of the front balcony. No renders of them yet, since I haven't even added the sides or handrails or anything, but they're there, I promise! =D I've also added the two lamps on the balcony, and I hope to build some of the ironwork between the ground and the balcony. I'll also have to run some tests to see which renders more quickly - truly three-dimensional modeled ironwork, or UV-mapped PNG images with transparent areas and invisible lights behind them to simulate ambient light from outside the ironwork. So much to do! =P
Thank you very much to everyone who contributed their surprisingly unanimous input! You made me feel a lot more confident as I made my decision, and I'm glad y'all apparently want the same choice I was leaning toward.
Actually, while we're on the topic of signs, I may as well explain how I go about making them. For some simple signs (an EXIT sign in the Indy queue comes to mind), all I really do is find a good photo of it and then slap that bit of the photo onto an object of the right size and shape. But for stuff like the entrance signs for the Dream Suite and Pirates, I prefer a somewhat more robust solution. The thing about photos is...they're accurate, but their resolution is limited. And in order for the image to appear sharp when it's rendered by Blender, it needs to have fairly high resolution. (Besides, you can get yucky stuff like noise and perspective distortion and objects in the way when you're dealing with photos.)
So my labor-intensive but oh-so-definitely-worth-it solution is to replicate the sign in Photoshop using appropriate fonts, vector techniques, and layer styles. I'm basically tracing over one or more perspective-corrected reference photos using the pen tool at first...then it's a matter of getting the colors and materials to look right. It takes a while - particularly tracing - but the result is a very sharp, clean image that can then be UV-mapped to the sign object in Blender. (Just doing a non-UV texture works too, since the sign object is typically flat and only one side of it shows, but that doesn't allow me to see the image previewed on the object as I navigate through the scene.)
As a couple of extra goodies to illustrate what I mean, here are my texture images for Pirates and the Dream Suite. In the case of these two signs, I've built the borders as separate, truly 3D objects that just take flat color materials; the UV-mapped sign object sits nestled within the frame, so you don't see that frame in the image. And notice how the corners don't show up on the final sign objects, since they have rounded shapes, so it doesn't matter what goes in the corners of the image. Also note how I have to simulate elements that would be truly three-dimensional in real life, such as raised lettering. I've also added a reflection to the Dream Suite image so that I don't have to increase render times by doing a truly reflective material. Finally, I want to point out that although the Dream Suite sign is way physically smaller than the Pirates sign, I've made the Dream Suite image significantly bigger so that the small Disneyland lettering is still sharp. If I wanted to get really fancy, I could put that in a smaller, higher-res image and let the rest of the image have a lower resolution, but I guess I'm just not that fancy.
Also, just so ya know, I've now added the curved steps on either side of the front balcony. No renders of them yet, since I haven't even added the sides or handrails or anything, but they're there, I promise! =D I've also added the two lamps on the balcony, and I hope to build some of the ironwork between the ground and the balcony. I'll also have to run some tests to see which renders more quickly - truly three-dimensional modeled ironwork, or UV-mapped PNG images with transparent areas and invisible lights behind them to simulate ambient light from outside the ironwork. So much to do! =P
Subscribe to:
Posts (Atom)
