I’m pleased to say that we are close to finish with the modeling phase. At this point there are only 22 components left (13 unique lego parts) to be modeled, reviewed and instanced. We still have to make all electric wires, rubber belts and stickers but it shouldn’t take much time. When it’s all done we’ll rig the whole thing and render a couple of final shots. Here’s one preview I did today (full rez here).

The image was stitch from 2 renders. Each baked roughly 6 hours.

As you can see theres lots of new, tricky parts. Most of them were done by Sergey Eizhvertinsh, so big kudos to Sergov for joining in! Piotr did few more as well, amazing quality as always. A couple of them were done by me too, but most of my attention was focused on assembling it all into one nice scene and coordinating our work.

As every long term project this one gives me a lot of satisfaction and opportunity to learn new stuff. Can’t wait to explore rigging in modo!

I’ll write more details about our workflow next time so stay tuned. In the meantime you can subscribe to speedies twitter to get notifications each time a new part is assembled.

P

Toast is a great modeler. Every piece was carefully measured to highest degree of accuracy (using real LEGO bricks) and poly modeled in 3ds max. You can find some of his fantastic work below.

I instanced his parts many many times over the scene. It’s a boring work, believe me . That’s how it looks at this point (click for highrez):

Some stats:

• 120 unique meshes
• 824 instances
• 308 011 raw faces (around 12M vertices render time)

Looks like we have around 70% of modeling done. Theres still lots of challenging bricks ahead though… I should have a bit more time to model some more this week and Sergo (another Speedies member) has prepared few more bricks that have to be reviewed, imported and instanced so there’s some progress.

I plan to rig the whole model once modeling phase is finished, but that’s future.

If you’re interested in how the scene looks real-time take a look at this video. It was recorded a week ago, so most of Toasted bricks are missing. You can watch it in HD if you wish, but you’ll have to go to vimeo site.

We have created a little tumblr blog too. Be sure to subscribe to RSS if you want to get news about this and other projects more frequently: http://speedies3d.tumblr.com/.

You can also talk with me or other guys directly on our IRC channel:

#Speedies at freenode.net (webchat link)

Finally, I may post some updates on my twitter from time to time too

Cheers.

Some wires:

I also set a test render with these bricks assembled.  It baked for 7 hours today. Here are two layer blending variations I like the most. Click for higher rez.

You can find other blends in links below:

alternative 1 , alternative 2

I plan to model all axles in a free time (parts ranging from 3704 to 3708, 3737 and 4519). In this case there will more work with instancing than modeling I guess… In the meantime here’s something special for all those interested in my workflow for lego modeling:

Cya!

Thanks for support,

Cheers

Pawel

It’s been a while. Here’s a little post I promised last time. T5 is a much bigger project comparing to previous Speedies entries and to be honest it’s far from finished at this point. In fact it’s on hold as I fail to find free time for some 3d fun these days and other group members seem to have the exact same problem.

I think I’ll just make a short series of posts showing you what T5 is all about, what has been done already and what’s yet to be done as I think it’s a high time to refresh things a bit here :]

Let’s start with a little introduction. Some of you may have noticed that I’m a bit nuts about LEGO modeling and rendering. I’ve been thinking about this project for a longer time really. T5 is about recreating official 8480 set – “FOS Light” Space Shuttle in 3d. I’m sure most of you have heard about this set as it’s pretty much a classic example of how brilliant LEGO can be. You can find some details about this thing here.

As you can see, it’s a very complex model. It contains around 185 unique pieces (1366 bricks in total). The plan is to polymodel and map all of them in detail so the bricks can be detached and used in different sets (so I model every bottom, stud, hole etc even if it’s covered by other piece). Once it’s all done the whole thing needs to be assembled and rigged. The last stage would be to render it.

So, this is a very ambitious project for me and a great opportunity to explore rigging tools in modo.

Some of you may have stumble upon these renders on RandomControl site. They asked me to share the scene as they wanted to test some core optimizations and upcoming modo plugin.

Current parts list:

78 parts waiting to be modeled (including 2 without a proper reference)
76 finished parts (including 3 that are waiting for a review)
30 parts in progress

More info in next post

Cheers

We planned to create a whole human figure but it quickly turned out that other Speedies members were pretty busy with their own stuff and couldn’t fully join me with this struggle. Eventually I went into „lazy mode” and modeled only main human head features, so I don’t have much to show you here aside of few wireframe images.

Anyway, I must admit that although my first head attempt was pretty sluggish, I’m quite satisfied with the final result. Here are some wireframes:

As you can see the topology is not so great in certain areas, cheeks for example, but overall shape is there.

If you’re interested in watching timelapse videos you can find some here: ear modeling, nose modeling, but that’s all I could upload as most of my recordings got corrupted due to some problems with CamStudio compatibility on Windows 7… Fortunately enough I solved it already and will upload more videos regarding Theme #5 soon. In fact I have few already compressed.

By the way, Theme #5 will be a bit different as it will be the first Speedies collaboration. I’ll reveal more details in a next post, but it looks like it will be much bigger than our previous projects. I might divide it into few „work in progress” posts with wireframes, videos and few test renders maybe, so stay tuned. Here’s a little teaser for you.

Thanks for stopping by.

Cheers!

Just wanted to write a quick note about stuff I’m working on right now. Sorry for a longer delay. I realize some of you are waiting for Fryrender tutorial I promised last time so I want to let you know that I’m still alive, working on it and on many other things.

Last months were pretty busy. I did few commercial projects,  reviewed some of my older scenes, trained organic modeling (my first ear ), started planning two collaborations and next Speedies theme (back to hard surface modeling) and raced a lot of laps in LFS . I also had a few short but lovely weekend trips with friends. Thanks guys!

Although Fryrender Scene Optimization tutorial is almost finished I’ll wait a bit with publication. I’ll send it to couple of guys and ask them to review it. I want to make sure all the technical stuff I’m writing about is true…

Besides this, I’m preparing more posts with Speedies updates, some stuff about productviz modeling, and another tutorial (short one) about texturing. Let’s say 33% of this stuff is done.

I’ve created a Facebook account lately, so if you’re registered there feel free to add me. If you’re not, I suggest you to try it as it’s a really cool site with lots of interesting features. It’s also a quick way to connect and share information. Let’s socialize!

Ok, so here are some of images I can show you:

About future plans, looks like there will be lots of modeling for both commercial and personal projects in upcoming weeks. I’ll try to update blog more frequently with some wires or renders maybe. Time will show.

Thanks for patience.

Cya

This time I wanted to make something less time consuming than previous entries. I picked Voyage concept by MILE, it looked simple enough, but once I started working on it… Well, let’s just say it was much too long. I was busy with other stuff, so it took me over a month to finish. When I grabbed all videos (this time nothing got corrupted) and stitched it into one big file it turned out to be around 8 hours. Not that bad, quicker than T1, but it’s still rather cold comfort.

Anyway, you can find the 50x version on Vimeo. As you may see on the video I blew the lid hinge at the first try… Modeling from photo reference is not the most accurate work flow around, but that’s all I’ve managed to find about this great phone concept. Even the exact size is a mystery…

Wires:

Some facts:

• Modeling time: ~7h 50m
• Raw video size: ~30,5 Gb, 1680×945 in 26 parts
• Speedies members were able to download x4 time lapse, 800×450 material in 8 parts
• Compressed video size: 410 Mb
• Polygons generated: 11 226 (11 194 quads)
• Efficiency: ~23,6 polygons/minute (T2: ~~54 p/min ; T1: ~~50 p/min)

Dividing polygon number by modeling time is not the most accurate way to measure your modeling speed I guess, but it gives a general clue about your productivity. Judging on efficiency I think it could be done twice as fast… Why is that? Mainly it’s because I was so bored with this project, but there was also some customization going on. It could slow things down as I wasn’t accustomed to the new environment. Besides that, I had to redo whole lid part…

Sooo, since it’s such a modeling fail let’s leave it behind and move to the render section

I did this one just to test materials. It’s heavily based on a photo by Takumi Ota (you can find it on AU design page). Initially I wasn’t planing to include it here but it turned out quite nice, so…

All these small details on gold material and keyboard are done with normal maps. They’re not perfect in every light condition but are much faster than displacement. I’m still learning how to create them well and how do they behave in Fry.

Oh, by the way if you’re from Japan ignore the keyboard, please The reference photo was too small to trace all characters, so I placed some random stuff there 8)

This time I forgot to note the render time. It baked roughly over night in this resolution and it was so clear that I had to add slight noise to make it look more realistic. Sp2 is fast, but speed is not everything. There are other problems with this version but I’ll save it for the next post

Next one baked for 10 hours in 1440×1800 even though the scene was totally unoptimized… There’s a lot of wasted samples here. Again, I’ll try to write a quick tutorial about optimizing Fry scenes in next post.

Cheers.

P.S Few days ago I got a mail from David Dopacio with a beautiful render of my chair model. I’ve included it in the original post. You can find it here. Thanks David!

Isn’t it a beautiful concept to make textures using DSLR cameras? It is, until you won’t place them on a big surfaces and hit tiling problems like uneven brightness, visible seams, color shifts etc. There are few tricks to deal with these problems though. One of them was recently showed here. It inspired me to write this post and show you how I prepare my textures before I make them tileable.

Personally I rarely use Photoshop to paint out tiling. Programs like Pixplant or Imagesynth let you make tileable textures automatically using seed images. Although it’s not a perfect solution for wood and other organics, it can save a lot of time. Basically, you can take few photos of a surface, import it and get a tileable output with just a few clicks. Theoretically it works with all “raw” photos but practically it’s better to invest more time and prepare them carefully before importing. Making them even in terms of color tone and overall brightness will pay off in the end.

I’d like to show you the way I equalize my seed images using layers in Photoshop. But firstly I have to admit that it’s based on this mini tutorial by Peter Guthrie. To be honest I just couldn’t grasp it fully so I started to mess around with layer blends and eventually developed my own technique. The goal was to minimize the amount of steps.

Step by Step

Bookmark Cgtextures.com if you haven’t done it yet

1. Download Image 1 (Large) from this page
2. Open it in Photoshop
3. Duplicate Background layer [ Ctrl-J ]
4. Open Filter > Blur > Average…
5. Double click Background layer to change it into movable layer.
6. Move it to the top of the stack, change it’s opacity to 50% [ 5 ] and blending mode to Linear Light
7. Open Filter > Other > High Pass… and play with Radius parameter to control the effect

Note: If you’re using the latest Photoshop version you may need to click “Show All Menu Items” in order to access Average and High Pass…

Radius of 100 works just fine in most situations but you might want to tweak it in the range of 50 – 150 depending on image resolution and size of detail.

Actions

I use this trick a lot, 90% of my textures are equalized and as you imagine, it’s not very handy to go through all these steps each time I edit a seed image. Turns out you can easily automate the process using Actions, so now it’s a single clicker (link fixed).

Examples

wood

copper

cardboard

Cheers

Even though Pedro has provided many reference pictures I was a bit afraid of this challenge. It’s the first time I model something related to guns actually.

I had some problems with setting reference properly but the AimPoint itself turned out to be quite easy to model. It was a quick session (~4h) so I wanted to add something more to make it look more complete. I picked a nice mounting and spent another 4 hours modeling it. Harder than I expected but still fun to do. Here’s the result.

Finally I did a couple of renders in Fry Sp1.

You can find 50x time lapse here. Unfortunately, 2 parts out of 10 got corrupted while recording and are not included. I’ve found the workaround for this problem (already added to recording tutorial) so hopefully it won’t happen with next recording session.

By the way, I’m working on Speedies T3 right now. This one goes extremely slow as I’m busy with other stuff, but I plan to finish it this week.

Oh, and the Speedies info section here… should be up very soon.

Cheers!