Forest Composition Progression

Well, I don't know why but I still can't seem to fix the flickering effect that's happening with the trees. I did manage to reduce it with a little motion blur and two resize nodes, but it's not completely eliminated. But just for the sake of progress, I'll post up what I have so far.

So let's start with the forest. I took 10 different images with alpha channel from the net as I mentioned earlier and used a particle emitter to replace each particle with each of these 10 trees and randomized the order so that it's more believable. Next, I duplicated and aligned the emitter to form a hilly forest look and created a 3D camera in the 3D space. I took a sky image which I had taken from my balcony, colour corrected it and placed it all the way behind the trees. If I'm not wrong, this technique is called multi-plane compositing or 3D compositing. Either way, it's still compositing :D

I animated the trees's spin variance in the emitter settings so that it would have this look of the trees swaying due to wind. For the foreground forest, I put in a higher value for the spin variance so that it would fake the effect of depth.

I went on to colour correction and adding fog based on the z-depth. I did not add depth-of-field because I felt that it wasn't necessary when it was such a long shot. So here's the before and after.

Before color correction / fog


After color correction / fog

With these, I was basically done, except the flickering effect as I mentioned. Took approximately half an hour to render without motion blur and 5 hours to render with motion blur! Here's the nodal workflow for the composition.

And here's the video. Simple camera move. My aim of this composition is to create depth and have the camera move through the space.

Here's a high resolution frame from the animation. Some would say that it isn't sharp, but that's the kind of effect I wanted it to have.

That's all for the forest composition, as far as my progress is concerned. I'm starting to get the hang of node-based compositing and I realise now why it's so much better than layer based compositing. For compositing, I would definitely use a program like Fusion or Combustion rather than After Effects, but for motion graphics, I would still stick to After Effects.

More later.

Forest

The idea was to create an entire forest using a few trees. I found about 10 different images of trees on the net which already had the alpha channel attached. I brought these into Fusion and attached them to a particle emitter so that each of these 10 trees would be multiplied and randomized to create a forest-like effect. I animated a 3D camera. Simple track. Here's what the node flow looks like:

My first render went wrong because the particles kept flickering, meaning the trees flickered. I still can't figure out what's the problem. So will do that and post the final animation as soon as I'm done. Meanwhile, here's the basic look.

I'll explain more details of how I made this when I'm done with the final render.

More Compositing in Fusion

Still in the process of playing around with Fusion. This time, I'm trying to create a matte painting of grass, trees, mountains and sky in Fusion and have the camera track in slightly. Well so far, I've just progressed with the grass part, which doesn't look any realistic at this point in time. Here's the node flow and a preview of what I've done so far:

Compositing is really fun if you come to think about it. All of a sudden, you tend to look things from a different perspective. Even when you step out of the house and you see a beautiful scenery, your mind tends to compose the image in front of you into layers and see the depth in it. Personally I think it's a damn cool thing, just that it's very tedious. Heck of a job. The only key trick to compositing is to get the lighting and colors right to make it look realistic/believable. Other than that, technical knowledge plays a pretty important part to help get things done.

Updated Hydrant

Thought I'd brush up my hydrant a little and share it here once more. Click on the image to expand.

I'll try to do this kind of thing for every piece of work I produce - the software's used and my initials.

More Fusion

I'm practically addicted to this program. Can't emphasize how awesome it is. I was practicing my compositing skills with different passes of the fire hydrant I created in Maya previously. For the previous composite, I did not change any color values. But this time, I decided to give it a warm look with a gradient background. So here's the node workflow of it:

And here's the final render of the composite:

Fusion Test

Hello everyone. I'm back again. I had been experimenting on Fusion for the past 5 hours approximately and I managed to come up with a composite of an earlier project that I had done. This time, I relied completely on the node-based workflow and created an entire composite out of it.

And here's the render of the final composite.

I just have one thing to say - Fusion is simply AWESOME! I had so much fun compositing in this program. Comparing it to a time-line based compositing workflow such as that in After Effects and Combustion, I had so much more flexibility using the node-based compositing workflow in Fusion. No doubt I'm feeling really at the top of the world that I'm starting to get the hang of this awesome program.

Think I'll catch some sleep now. It's 4.45 in the morning (=

Mini Personal Project - Hydrant in Live Action

Okay, so this blog has been pretty dead recently, so I'd thought I'd update it with a little bit of what exactly I've been doing recently.

Okay, so mostly I've been doing research lately on compositing. A lot of it. I've been exploring Combustion, which I personally find to be much better than After Effects in terms of compositing. However, I'm still not very comfortable with it, so I'm still going on about it. I'm also partly exploring Fusion, which is a hell of a complicated node-based program. Lots of patience required. Other thank compositing, I've been researching on different lighting and rendering techniques using Maya and mental ray. All of this research will be applied on a short mini personal project I'm working on.

This personal project of mine involves the latest fire hydrant that I've modeled. What I'm aiming for this project is seamless compositing of my CG hydrant into the live action background footage. So I went downstairs with my video camera and still camera and did a little bit of shooting.

The above represents a still frame from the live action shot. Basically, the hydrant will be placed at the empty area on the frame. After shooting the hand-held camera shot, I used my still camera and took about 7 photographs of different exposure levels of the surrounding area so that I could create a HDR image.

After the shooting was done, I converted my footage into a sequence of frames and had them tracked using boujou. Awesome program to get accurate 3D tracks. I also created my HDR image using Photoshop. Once that was done, I got into Maya and started creating the geometry to match up the orientation of my hand-held camera with Maya's 3D camera.

Once the geometry had been matched up, I imported my hydrant and placed it in the scene and scaled it accordingly.

This is where I've stopped. Basically the whole point of this personal project is to practice and test out my lighting, rendering and compositing abilities to get as perfect a seamless composition as possible. Once I'm done rendering several passes such as diffuse, occlusion, specular, etc., I would composite these elements in Combustion and Fusion and try to get both the end products to look as similar as possible.

So that's the update so far. More later.

Textured Hydrant

I laid out the UVs for the hydrant. I still don't know much bout UV, but I decided to apply a simple cylindrical map. Then, I added a fractal map as a bump map and assigned the blinn shader to the hydrant. I rendered in three passes - diffuse, specular and occlusion, and composited everything in photoshop. This is what it looks like:

Fire Hydrant

It's not that I get side-tracked a lot. This time, it was during class when Greg and I decided to model a fire hydrant as soon as possible. Why a fire hydrant of all things? Because we wanted to model something non-organic. Secondly, my team, Crewsade, is working on this project which revolves heavily around fire fighting and we definitely need fire fighting related assets like fire hydrant. So we decided to model it.

I started out with a reference image below. It's an American fire hydrant. Basically what I loved about this model is the small details that make the hydrant look really interesting.

However, for my own model, I decided to go for a different approach. I modeled using a cylinder for a base shape with very low resolution on it. I sculpted the hydrant to match my desired look. The challenge between Greg and I didn't last that long because class got over, so I just continued at home. I added in the details such as the repetitive dents on the metal and the bolts and nuts.

Once the modeling was complete, I was too lazy to texture it properly. So for the time being, I added a plane for the ground and rendered an ambient occlusion pass with final gather. Here's what it looks like.

One thing I decided while modeling this hydrant - everyone of my work will have my initials somewhere on it. Not like a typical signature, but instead the words "Digimator's Renders" or "DR" on it. It'll be like my own way of marking my work out (= So here are the close-ups of my initials on the hydrant.

I'm pretty excited about this model. I plan to texture it, light it and render it semi-photorealistic pretty soon when I get more time. Can't wait to see what the finished product looks like =D

Oh yes, and I haven't forgotten about my car. I'm still researching on the topic and different methods and workflows for modeling. Stay tuned :)

Ambient Occlusion Animation

The animation is up on youtube.

Click here to watch it.

No big deal actually. Just a simple camera pan around the text, showing the shadows.

Ambient Occlusion

I actually wanted to get started on the car, but got sidetracked because I got curious about ambient occlusion all of a sudden. So I decided to research on what exactly ambient occlusion is and played around in Maya by doing a simple logo animation of Digimator's Renders. Yes, it's pretty lame right now, but I guess it can do for a test render.

Render time was pretty expensive. I turned on final gather and global illumination. Approximately 5 minutes per frame. The animation is being rendered at the moment. I'll upload as soon as it's done.

Truthfully, my research on ambient occlusion seemed pretty complex to me. However, I did get one thing right out of it - photorealism. Ambient occlusion can be pretty useful when doing visual effects - CG compositing with live action. It plays a great part in photorealism in that it gives very soft, diffused shadows and very dark shading when corners meet. From what I understand through my experimentation, I think that ambient occlusion is another way of defining soft, diffused, realistic shadows. I may be wrong. I definitely plan to explore in depth on this matter, and how to use it through Pixar's RenderMan.

BMW Update - Rendering & Reflecting

I was doing a little bit research on texturing cars and I decided to do a few test renders on my BMW Roadster. Here are a few of them, which are compared to the original car.

Basically what I did was very simple. I assigned a mental ray car paint shader to the model and tweaked the shader a little to create a grey car. I turned on image-based lighting. Since I did not have a HDRI image, I used a jpeg instead and adjusted the color gain so that it would emit enough light. I added a directional light for some extra lighting. Finally, I turned on final gather and global illumination and rendered the image through mental ray.

Some close-ups:

I am going to start all over again. Even though a lot of time and effort was spent on this, there are many obvious flaws which are very difficult to correct at this stage. The first mistake I realised is that I jumped into details very quickly into the early stage of modeling. Maybe I was too excited to see the final result like most beginners. I did the edges too quickly and extruded the door sooner than expected to create that little bend. I should have blocked out the main body of the car first before adding too many edge loops to create the detail - which explains why the car looks dented at many points due to meddling around with the vertices. Secondly, I should have kept the car in separate pieces - the front part, the door, rear etc should have been separate pieces on their own so that I could create those panels perfectly without getting too much into extruding the faces to create those small ridges between each panel.

Nonetheless, I'm quite happy with the result here. I did get the edges where I wanted them to be. The only reason why I am starting all over is because I want to avoid these mistakes that I made and follow a proper workflow to create this car. After all, it's my favourite car and every effort should be put in to make it as perfect as possible! (=

Demoreel 2008

As I had promised, I'm finally done with my demoreel. Actually I was done quite some time back, approximately a week ago, but I had issues compressing it for web streaming. Well anyway, here it is:

Demoreel 2008

Do remember to turn on your speakers (=

It's the compilation of my latest works. Longest one I've done so far. Compiled entirely in After Effects. Do feel free to comment on it.

I'd like to thank Tad Leckman from Lucasfilm Animation Singapore for guiding me on improving my demoreel during the consultation session at NAFA after their talk on how to make a good demoreel.

Enjoy! =)

P.S.

For those who can't access the youtube version video, here's the blog version:




Youtube version would still be a better choice though.

Enjoy once again.

Subsurface Scattering Experiment

I keep on hearing this term and after Alan explained to me about what it was, I decided to play around with it.

So I built a simple model, inspired by the Hulk, his upper half and rendered it with subsurface scattering through mental ray. This is what it looks like.

The whole point of this subsurface scattering technology is to get realistic looking skin. I don't quite know whether I got there or not. I applied a dielectric material to the ground so that it would give a nice reflection and turned on final gathering. I used two lights in the scene. One spotlight for the backlight and a directional light for the fill light.

I then twitched around with the model to exaggerate the muscles using the sculpt geometry tool. Once I was happy with the model, I rigged it and skinned it very roughly and adjusted the arms to give it a pose and rendered it again.

I was quite happy with this. I've never modeled a human figure before, and although the anatomy is wrong here, I still like the fact that it's over-exaggerated. Perhaps the skin might need to look less plastic like.

Next, I wanted to make it look like a stone sculpture. I carried on playing with the texture.

Was quite happy with this. It's definitely inspiring me to make an exaggerated anatomical human-like creature.

No updates about the car yet. Haven't gotten the time yet. More later.

Back Part Started

Been very very busy with school work - crafting resume, cover letter, shot breakdown and creating my demo reel (which should be up here and on youtube pretty soon). So couldn't do much on the car. Here are the updates.

As usual, edge flow is always a problem. I messed it up a little, but fixed it back. I'm still thinking whether I should connect the back part to the front. Maybe it would be a lot easier for the entire modeling process.

More later.