At long last, the project is almost over. Less than two days away as of this writing. After delivering the final film a couple of weeks ago, I’ve spent my time relaxing, creating a short “making of” with the rest of Green Spill, writing stuff on this blog and a self assessment document. The self assessment document isn’t quite finished yet. I’ve written most of it in short form which I suspect only I would understand, so I have a feeling I should go back and do a rewrite or three. Also yet to be done is to update the DVD cover and label. We need one version for the “making of” we’re handing in on wednesday, and another version for the team: The version destined to collect dust on a shelf somewhere, but which will be taken out and looked at every few years. The very special Green Spill team member version. With any luck, I’ll be on that right after finishing this post, and the only thing I’ll have to worry about tomorrow is straightening out the self assessment document to something resembling the language of english.

This being the last post I write about the Student Film Project at Noroff, 2009 (at least I think it is), I guess the time has come for looking back and reflecting upon the last five months of work. Don’t worry, I’ll be quick. Or as quick as I can be. Which can be pretty slow sometimes, but I digress.

Looking back, it has certainly been an interesting five months. We’ve gone back and fourth on story, had our moments of inspiration and desperation, experienced conflict and unity and last, but not least created a film!

The film itself turned out OK, though I certainly think there is room for improvement. Story, continuity, animation, graphics, simulations, compositing and editing spring to mind as points where we could have improved given more time or by making better use of the time given. Having worked on the film for five months and knowing about every fault from the tiniest to the most humongous, I might just be extremely fed up though

On the whole, I’m one experience richer what film making is concerned, have learned a lot, both about team work, planning and skills, and though it has been trying at times, it’s an experience I wouldn’t be without. I think all there is left is to thank my team mates, Aleksander, Andreas and Jamie and wish them, my teacher, Paul, and the rest of 3D Film Production in Oslo a very happy holiday

Now onto those DVD-covers…

Having finished rendering and vfx shots, and with a little more than a week to deadline, compositing was next on the list. Beforehand, I had estimated compositing would take a day or two. I had already test comped nearly one third of the shots and thought I could use those as templates for the rest. The only thing I hadn’t done before in any shot was thinking about the suns position and the movement of the clouds.

When applying math however, it turned out I would have on average 11 minutes to complete a shot, rendering time included, if I were to do all shots in one day and have time to eat and sleep too. Considering I would have to do some rotoscoping on most shots, and the longer shots could take up to 12 minutes to render, I set my aim to finishing in four days in stead. This would give me approximately 45 minutes per shot.

The process went fine, though it was rather tedious and repetitive. While working on the first few shots, I realized the ground had a quite homogenous yellow colour that looked pretty boring. Seeking to resolve this, I thought I’d introduce cloud shadows on the ground to break things up a bit. Shadows moving across the landscape would also slightly liven up the background. I had seen Gareth Edwards do this on Attila the Hun and thought it had a nice effect. Unlike Edwards however, I didn’t have the render passes and with no time left for going back into 3ds max, I decided to fake some. My approach was to use the noise map in Nuke combined with a corner pin to make the far away shadows appear smaller than those closer to the camera. This worked well as long as everything stayed nice and flat, but the effect was ruined if something that wasn’t flat and level with the ground entered the frame. Such as a prop or a character. As a fix, I had paint out the noise map where it happened to intersect with a character or prop. I thought nothing of it for the first 10 shots or so, but then realized there were 54 more shots to rotoscope in the same way. Luckily, I soon discovered a better way that worked most of the time while watching Greg Astles talk about Nuke on fxguidetv #059. By using the ZBlur (Depth Of Field) tool in an unconventional way and leaving the Focus Plane feature on, I could zoom in on a particular “slice” of the z-depth map I had already rendered and get a decent matte from it. I later found the ZSlice tool which I guess does the same thing, only simpler to use.

I ended up compositing the film in four days, though I had one day off in between, and am reasonably happy with most of the shots. Some shots have a look that is a little too different from the rest, but I really didn’t have that much time to stop and reflect when compositing 16 shots a day.

With less than three weeks left of the project, animation was finished to a point where I could start adding fire and smoke to finished shots. Having used FumeFX in the past, it became the natural choice now.

I anticipated this to be a fairly simple task, but didn’t have to work long before it became clear that I had greatly overestimated the ability of 3ds max plugins to work properly together. Having successfully used FumeFX with V-Ray before, it never occurred to me that one tiny change would throw the whole thing off balance. It did though. As it turns out, FumeFX works perfectly with V-Ray as long as you don’t use the V-Ray daylight system. Light from the V-Ray daylight system is almost infinitely brighter than normal lighting in 3ds max, and the only thing keeping the images from turning out simply white is the VRayPhysicalCam which dials the brightness back down again. As to why the daylight system behaves in this peculiar way, I have no idea. The problem surfaces when you wish to put a flame from FumeFX in this environment. FumeFX uses normal brightness values for its flames, and when the VRayPhysicalCam reduces the brightness by a factor of a thousand or so, the FumeFX flame is simply too weak to register in comparison and ends up invisible. I spent a day or two, looking for a solution, but didn’t find one and was left with two choices: Getting rid of the daylight system, relight the scene and scrap 30-40 shots already rendered, or render the FumeFX elements alone and composite them over the rendered shots afterwards. Being pressed for time, the compositing route was the only option if we were to finish on time. There are a fair few advantages to this, such as being able to adjust the brightness and colour of the FumeFX elements without affecting the rest of the image and being able to adjust fire and smoke separately, but it also comes at a prize as you no longer get out of the box integrated smoke shadows. The case with the V-Ray daylight system this is particularly irksome as FumeFX doesn’t even see the sun as a light source for generating shadows. My solution was to create a Direct light in the same position as the sun and using that instead, but it didn’t always yield the best results.

Having decided on a workflow, I started work and the rest of the process was fortunately pretty straight forward. I finished the FumeFX shots in a little more than a week while rendering and compositing other shots at the same time. Having finished, it was time to proceed to the final compositing.

Well, technically it’s been finished for a week now, I’ve just been too fed up to blog about it until now. The past month has been incredibly busy and seeing as my field, post production, comes last (who would have thought?), I haven’t been sleeping that much lately. Lack of sleep coupled with missed deadlines and frustration levels through the roof generally meant that keeping the blog updated became one thing too many. Taking an hour or two out of the on average five hours of sleep I got every night to blog didn’t feel particularly inviting. And if I had taken the time to do it anyway, it would probably have made for rather unpleasant reading.

So I’m picking up the thread a week later, now having had a few nights normal sleep and some time to relax and think about things other than the film. I’ll be back with a few posts about the past month and reflections on the finished film in the next few days.

So we’re finally at the animation stage. Or have been for the past few days. To get the most out of the remaining time, Andreas, Aleksander and Jamie are chipping away on the shots while I render them out and take them through compositing.

As I’m the last person to touch the footage before editing, I’ve spent the past few days trying out different “looks” for the final film, screened them for the rest of the team and adjusted things based on the feedback. We finally decided on the look in the video above. I then created a Nuke template that I can use to quickly apply the same look to the rest of the shots.

The only thing remaining for actually pushing out finished shots is to work out the wind direction and position of the sun for each clip so that the clouds move properly etc.

To stay sane, we’ve created an Excel document to keep track of what everyone is doing.

We’re also planning to have a bit more convincing smoke and fire effects this time around, so I will add smoke and fire to certain shots when they come in. For this reason, I’m currently spending my time learning the 3ds max plugin FumeFX so we’ll be able to produce convincing fireballs, smoke trails and explosions. Allan McKay has some good tutorials on the subject, so I’m looking through those. Here’s hoping V-Ray and FumeFX will play nice when it comes to render passes…

Compositing in After Effects is in many ways very similar to working in Photoshop. If you are familiar with layers, masks, blending modes and/or filters, compositing in After Effects should be a breeze.

Go to Part 1

5. Importing footage

Open After Effects. Press Ctrl+I for Windows or Cmd+I on the Mac to import the render passes. Navigate to correct folder and select all files by pressing Ctrl+A/Cmd+A. Make sure to check the box called Multiple Sequences. Then press Open.

After Effects is now going to ask you whether the Targa files are premultiplied or not. As we checked the premultiplied box when outputting files from max, they are indeed. We didn’t set any background color in 3ds max and so it should be black which is the default. As a result, we’ll choose Premultiplied – Matted With Color and make sure that color is set to back which is the default in After Effects.

As you are importing four image sequences, you’ll be asked the same question four times. Answer Premultiplied – Matted With Color every time.

6. Layers and blending modes

Now that the footage has been added, you need to drag it down to the time line. Start with your main pass, in my case city{0000-0100} and scrub to a place in the time line where you are able to see both the city and the sky.

To add a bit more light and a hint of global illumination, we’ll add the occlusion pass next. This pass goes above the main pass.

To see the main pass through the occlusion pass, we need to change the blending mode for the occlusion pass from Normal to Soft Light.

You might need to press the middle button in the lower left corner to get to the blending mode options.

To make the scene less gray and boring-looking, we’ll give the scene a yellow tint as if sunlight was illuminating the city. Right-click below the two render passes you added, choose New and then Solid.

Then, under Solid Settings, give the solid a light, yellow color.

Change the blending mode for the solid to Multiply.

As you can see, this gives us a slight problem as the sky just turned yellow too. This is because unlike the rendered passes, the solid does not come with a built-in alpha. To add one and get rid of the yellowness, first drag the matte render pass into the time line. Drag it to the bottom of the stack.

This seemingly cleared up the matter at once as the sky instantly went back to black. However, this only happened because the underlying matte is black in the area representing the sky and that anything (including light yellow) multiplied by black becomes black. To make the sky area of the solid truly transparent, you need to right-click the solid, choose Effect, Channel and then Set Matte.

In the Take Matte From Layer drop down, select the matte which will be the fourth option. As the matte doesn’t come with an alpha channel, you have to change the Use For Matte option from Alpha Channel to either Red Channel, Green Channel or Blue Channel. As the matte is grayscale, the red, green and blue channel contains the exact same data and it doesn’t matter which one you choose.

The solid is now transparent. As we’ll be needing to apply the matte to other layers later, select the matte in the Effect Controls window and press Ctrl+C/Cmd+C to copy it. Next up is the z depth pass. This can be used for two purposes, but here we’ll use it to decrease contrast further back to put some “air” between the foreground and the background. Drag the z depth pass to the top of the time line.

As we wish to decrease the contrast more the further away you get, we want the lightest areas in the background. To achieve this, we’ll invert the layer by right-clicking it, choosing Effect, Channel and Invert.

Then set the blending mode to Add.

We have now successfully put some “air” between the foreground and the background, but in reality, this “air” isn’t white on a sunny day, but the color of the sky. We’ll apply a color correction to the z depth pass by right-clicking it again and choosing Effect, Color Correction and Hue/Saturation.

Check the box called Colorize and then turn the Colorize Hue knob to somewhere round -160.

Also, the sky has turned white, due to the fact that the z depth pass does not contain an alpha channel. With the z depth pass selected in the time line, highlight the Effect Controls panel above and press Ctrl+V/Cmd+V to paste in the matte from earlier. The Take Matte From Layer drop down resets for some reason, so we have to select our matte layer again.

7. Sky replacement

Now that we have a basic composite for the buildings, we’ll proceed with replacing the sky. First step is to find a suitable image. Personally, I went over to CG Textures and downloaded one of their Skies 360 images. Import the image by pressing Ctrl+I/Cmd+I and pointing to it on your hard drive.

Next, drag the new sky layer into the composite, directly underneath your main layer.

Right away, the sky looks too large and rather boring so we’ll try making it a bit more interesting. First, reduce its size by expanding the layer (press the little triangle) and scale it down to around 40% under Transform.

In the viewer, drag the sky layer upwards to match the horizon in the cityscape.

Select the sky layer and press Ctrl+D/Cmd+D to duplicate it. Rename the layer to something like Sky Flipped. To do this you select the layer on the time line and hit Return on the keyboard to rename it. As you might have figured out by now, we’ll be flipping this layer. We’re doing this to conceal the fact that we’re using the same sky map twice. Expand the layer again, uncheck the little link symbol next to Scale and set the first value to -40% instead of 40%. Then check the link symbol again and in the viewer move the sky so the cloud formations doesn’t overlap with the same ones in the layer underneath.

Now for the fun part: Still using the Sky Flipped layer, play around with different blending modes until you get a dramatic effect. I ended up with Linear Light. Don’t think about matching the color to the rest of the image yet, we’ll look at that next. You now have two sky layers you can animate at different speeds, making it look like some clouds are moving faster than others across the screen. Don’t animate anything just yet though.

To make the sky and the 3d clash less, we’ll create what’s called an Adjustment Layer. An Adjustment Layer doesn’t contain graphics, but will let you apply effects to it that also applies to all layers below it. To create one, right-click any empty space in the time line and choose New and Adjustment Layer.

In this case, we only want the Adjustment Layer to affect the sky, so we will drag it to right above our two sky layers. If you like, you can rename the layer to something like Sky Adjustments. Now that we have the Sky Adjustments layer in place, right-click it, choose Effect, Color Correction and Hue/Saturation.

Now, dial Master Hue back to round -10 and Master Saturation to about -50 and the sky and city should begin to match each other.

8. Tracking

The sky replacement is now almost finished, but one major thing is missing. While the rendered footage is constantly moving, the sky maps are stationary, making the composite painfully obvious as soon as you hit the play button. In other words, we’ll have to animate the sky, making it look like it follows the city horizon. Doing this by hand can be very time consuming and painful, though luckily there is a better way called tracking. When tracking a shot, you’re basically making After Effects watch a few selected areas of the image, frame by frame, and recording their movements. You can then apply the same movement to other objects in the scene. Areas used for tracking are preferably points of high contrast that don’t deform too much during the clip, making it easy for After Effects to recognize through a long sequence of frames. In this respect, the shot we’ve just created is rather difficult to track. Nearly every part of the rendered sequence deforms over time. In addition the sky needs to move with the horizon, confining the area of tracking to that fine line which is almost constantly deforming. In other words, tracking this shot, although possible, will probably be difficult and time consuming.

Luckily, we’re talking about a 3D render here, and so we have the means to fix this pretty easily. Go back to 3ds max, create two pyramid shapes and place them close to the horizon, looking through the camera. Make sure to space them a little apart. Now, hide all other geometry in your scene, give the pyramids a white, self-illuminated texture and render the sequence against a black background. Instant tracking markers!

This makes tracking a breeze. Import the pyramid pass and drag it to the bottom of the time line.

We also need something to apply the motion to. For this we’ll be using what’s called a Null Object. Basically an invisible object that can store information about position, rotation and scale. Rather like a Dummy in 3ds max. To add one, right click an empty spot on the time line, choose New and Null Object.

With the Null in place, right-click the pyramid layer and choose Track Motion. A panel named Tracker Controls will appear on the lower right of your screen.

For this scene we need to track both Position, Rotation and Scale so make sure all three boxes are checked. As it’s hard tracking rotation and scale with only one Track Point, an additional point will appear once you check them.

Click the button called Edit Target and make sure it is set to our Null Object. This will make sure the motion data will be transferred to the Null Object when racking is finished.

Now we need to decide whether we are going to track forwards or backwards. In our case, the tracking points (pyramids) are actually out of frame at the beginning of the clip, so we’ll track backwards. That means scrubbing the time slider to the last frame.

Having chosen a frame, we need to align the Track Points with the pyramids. Click and drag one to the lowest corner of each pyramid. We’re choosing the lowest points as they are the last parts of the pyramids to exit the frame before they disappear completely, which gives us another couple of frames of tracking.

Now press the Analyze Backwards button and watch After Effects tracking the scene. As you can see, the pyramids disappear after a while, making the Track Points stop, but after that point we don’t see the horizon anymore and as the point here is to track the sky, that’s perfectly acceptable.

Having finished the tracking, press the Apply button to transfer the position, rotation and scale information to our Null Object. Dimensions should be both X and Y.

Try playing the animation to check whether the Null is now fixed to the horizon.

The final piece of the tracking puzzle is to scrub to the frame where you aligned the sky horizon with the city horizon and parent the the sky layers to the Null once there. Parenting in After Effects is similar to linking in 3ds max, meaning the parented layers will follow their parent. To parent both sky layers to the Null, select them and drag the little spiral icon on one of the layers to the Null Object.

Play the animation again and make sure the sky stays within the frame at all times. If it doesn’t – move the sky layers around in the viewer until they do. Congratulations! You have successfully tracked a shot in After Effects! You can now go ahead and animate the sky layers if you want.

9. Blur, grain and pre-comps

We’re definitely getting somewhere now and to round off the shot we’ll add some depth of field, blur and a bit of grain. All three effects helps us hide the fact that the buildings are a bit low-poly to be in the foreground and that the sky and city doesn’t blend perfectly yet.

9.1 Depth of field

As depth of field is an effect we want applied to the whole image, we’ll first create a new Adjustment Layer and make it the top layer. I’ll name it Depth of field. We’ll also drag in a new z depth layer and make it the bottom layer. Having done that, right-click the newly created Adjustment Layer, choose Effect, Blur & Sharpen and Lens Blur.

The image instantly blurs all over. In the Depth Map Layer select the bottom z depth layer. Also check Repeat Edge Pixels to get rid of the dark border surrounding the frame.

Use Blur Focal Distance to focus and Iris Radius to control how much blur you apply. In this example I set Blur Focal Distance to 215 and Iris Radius to 25.

This leaves us with one problem though, as the sky gets blurred way too much. To fix this, we have to tweak the z depth layer we just added a little bit. Select the bottom z depth layer and in the menu bar go to Layer and Pre-Compose.

This creates a new composition within our main one, meaning we will be able to output the new composition as a new, modified z depth layer. Name the new composition Z, move all attributes to the new composition and open it.

As you can see in the new composition, the sky area of the z depth is represented as pitch black, making the Lens Blur effect read the sky as really, really far away and that it therefore should be really, really blurred. We’ll change the color to something brighter, making Lens Blur go more easy on the sky.

Create a solid by right-clicking in the timeline and choosing New and Solid. Make the color light gray this time. in this tutorial, I’m going for #8A8A8A. The solid goes below the z depth layer.

Drag in a new matte from the Project tab and make it the bottom layer.

Right-click on the z depth layer, choose Effect, Channel and Set Matte. Set Take Matte From Layer to the matte and set User For Matte to Red Channel, Green Channel or Blue Channel. Now we have a z depth map with a lighter background.

Back in the main composition, you’ll notice the sky is significantly less blurred.

9.2 Blur

I would recommend adding a small amount of blur to the composition as 3d renders often gets too crisp. To do this, simply add another Adjustment Layer, call it Blur and add the effect called Fast Blur. Set the Blurriness to somewhere between 0,5 and 2 and make sure to Repeat Edge Pixels.

9.3 Grain

Finally, if you wish to give the shot a more filmic look, you might consider adding some grain as well. Create yet another Adjustment Layer, call it Grain and add the effect called Add Grain. There are a lot of settings to play with here, but I ended up with the Preset called Eastman EXR 50D (5245) and decreasing the Intensity to 0.9. I also set the Viewing Mode to Final  Output.

I guess this is all for now. I know there are still a ton of things I could have gone in more detail about or explained better, but I had to stop somewhere If you’re stuck, want something explained better or just want to tell me how plain wrong I am, feel free to leave a comment and I’ll answer you as soon as I can.

Go to Part 1

In this tutorial I’ll take you through the process of quickly creating a textured city in 3d studio max 2009 and how you, through a few extra render passes, can go from a quite crappy render to something half-way decent using After Effects CS3. If all goes well, we’ll end up with a finished shot not too far from the left side of clip above. In part 1 I’ll cover the creation of the city in max and in part 2 we’ll jump over to After Effects for the compositing.

Go to Part 2

1. Creating buildings

Create a few buildings in 3ds max or use the max file provided in this tutorial.

Lastly, you’ll end up with a total building count round 3000 so low poly is probably a good idea if you don’t have a very fast system.

2. Particle Flow

Create a plane and name it CityEmitter. The buildings you created in the previous step will be distributed across this plane.

Next, go to the Create tab, choose Particle Systems from the drop-down menu and choose PF Source.

Place the PF Source anywhere in your scene. The position is not important.

Under Emission, set Viewport Quantity Multiplier to 100% so you’ll be able to see all buildings in the viewport.

Then, under Setup, press the Particle View button. When you open Particle View you’ll get what’s called a Standard Flow which is basically a list of controllers that control birth, position, speed, rotation and shape of the particles. You also have control over how the particles will be displayed in the viewport and during render. Near the bottom is a list over all available controllers.

We’ll now start by adding one of our buildings to the scene:

2.1 Birth

As we don’t want buildings to pop out of the ground during animation, all particles should appear simultaneously. We’ll therefore set Emit Start and Emit Stop to the same frame. I’ll set both to -1 to make sure the buildings are in place before animation starts at frame 0. In my case I want 250 buildings and consequently set the Amount to 250.

2.2 Position

As you can see, this gave us 250 particles emitting from the PF Source icon.

To attach the particles to the plane instead, we’ll need to replace Position Icon in the controllers list with something else called Position Object. To do this, find the Position Object controller in the list of available controllers and drag it on top of the Position Icon controller in the view above. Let go when a red line appears over the Position Icon, meaning it’ll be replaced by what you’re dragging in.

Having selected the newly created Position Object, you can now press the button called By List under Emitter Objects which brings up a window where yo can select an Emitter Object. Choose the plane you created earlier called CityEmitter.

This randomly distributes the particles across the plane.

However, if you scrub through the time line, you’ll notice that the particles falls away from the plane which to state it mildly is less than ideal. To fix this, simply check the box called Lock On Emitter and the particles will stay in position.

This is all good of course, but at this point the particles are placed completely random and in a city with streets, this isn’t very realistic. To fix this, you first need to create a black & white material for the CityEmitter. Light areas will receive the highest density of particles while completely dark areas will receive none. In the example, I’ve combined the procedural textures Tiles and Gradient Ramp to create a texture, but you could also paint one in Photoshop or similar to get a more organic distribution map.

Back in Particle View and still under Position Object, scroll down to Location and check the box called Density By Material. The particles are now concentrated in the lighter areas.

To make sure the buildings don’t stand on top of each other also check Separation and play with the Distance until the particles have some space between them.

2.3 Speed

Buildings really shouldn’t move around that much, so the speed controller is useless here. Right click it and choose Delete.

2.4 Shape

Now we’re going to replace the rather boring-looking ticks with the actual building. Clicking on the Shape controller doesn’t give us a whole lot of options so we clearly need another one: Shape Instance. Replace Shape with Shape Instance the same way you replaced Position Icon above.

Then, with Shape Instance selected, press the button called None under Particle Geometry Object and choose a building by pressing H on the keyboard. I’ll pick my SkyScraper object.

Aaaand, voila! Nothing happened! This is because the Display settings below are still set to Ticks. Quickly select Display in the list and choose Geometry from the Type drop down menu. We now have buildings, though their rotation might be a tiny bit off. We’ll look into that soon.

Go back to Shape Instance and adjust Scale and Variation according to your preferences. I’ll keep Scale at 100% and set Variation to 25% to vary the landscape a little bit.

2.5 Rotation

Select the Rotation option from the list. As you can see, it’s set to Random 3D which explains why practically no building is standing upright. From the drop down menu, choose Random Horizontal to only make the buildings rotate around their Z axis.

If you don’t wish the buildings to be rotated at all, right-click the Rotation controller in the list and choose Delete.

2.6 Adding more buildings

All that remains now is to add the rest of the buildings. Fortunately, this is a very simple process. In Particle View, select the box called Event01, right-click it and choose Copy. Then right-click somewhere else in the same window and choose Paste.

Then hook the new Event to the PF Source.

Now, you only need to change a few parameters to have your new building in place. First go to Birth and set the number of buildings you want. This time I want 750.

Then go to Position Object to change the position of the new building so they won’t be positioned in the same spots as the previous ones. Scroll down to the bottom of the list and under Uniqueness press the New button to generate a random Seed number.

Finally, under Shape Instance, choose a new building by pressing the button under Particle Geometry Object, pressing H on the keyboard and selecting a new object from the list. I’ll choose Rubble. Then play around with Scale and Variation once more until you’re satisfied.

Repeat paragraph 2.6 until all buildings are scattered on the emitter. I chose 500 OfficeBuildings and 50 IntactSkyScrapers.

3. Finishing the scene

As you probably don’t want your ground texture to look like the distribution map, you need to hide the CityEmitter and use a plane with another texture instead. The easiest way is to select the CityEmitter, press Ctrl+V to duplicate it and choose Instance. Name the new plane Ground.

Then press H on the keyboard, choose CityEmitter and press OK. Right-click the plane and choose Hide Selection. You are now free to apply which ever texture you like to the Ground plane. I chose my very simple Rubble texture.

Now you just need to add some basic lighting and a moving camera and you’re ready to proceed to rendering. For lighting I suggest a single direct light with a mental ray Shadow Map for shadows.

4. Rendering

As mentioned above, we’ll use mental ray for this tutorial, so the first thing to do is to enable the mental ray Renderer. Press F10 on the keyboard to open the Render Setup window and under the Common tab, scroll down to the bottom to the Assign Renderer section and press the button next to Production. Choose mental ray Renderer from the list. Next, scroll up again to set your resolution. In this example I’m using 640*360, but you are of course free to choose whatever resolution fits your needs. Next, set your frame range, 0-100 in my example, and choose a place for saving your rendered files. I suggest creating a render folder as you’ll end up with quite a few files. The file format we’ll use is Targa Image File with the following settings:

Still in the Render Setup window, proceed to the Indirect Illumination tab and make sure Enable Final Gather, Caustics and Global Illumination are all turned off.

Then proceed to the Render Elements tab. You’ll need three additional render passes for this tutorial: A Matte to separate the city from the sky so we can do an effective sky replacement, an Occlusion Pass to fake global illumination and a Z Depth Pass to be able to add an atmospheric effect and some depth of field.

4.1 Occlusion Pass

Press the Add … button and choose mr Shader Element from the list.

Name the pass Occlusion and under Parameters press the button next to Shader to add a new shader. Choose Ambient/Reflective Occlusion from the list. There are a lot of settings to play with to tweak the Ambient/Reflective Occlusion shader, but the standard settings, although a little rough, will do just fine for this tutorial.

This basically renders a pass of the scene where this texture is applied to all geometry.

4.2 Z Depth pass

Press the Add … button again and this time choose Z Depth.

The Z Depth pass produces an image where the geometry close to the camera is displayed in white while geometry further away gets darker and darker until it reaches a cut-off point and everything behind it is rendered as black.

Under Z Depth Element Parameters you need to set a cut-off point for where the image goes entirely white or entirely black. In the example scene a Z Min value of 0 and a Z Max value of 2000 works out quite nice as the city plane is 2000 units wide and the camera is looking at the city from one end of the plane.

This might need some tweaking though, so I suggest doing a couple of test renders of still frames before rendering the final sequence.

4.3 Matte

Press the Add … button for the third time and this time choose Matte.

A matte lets you isolate an object in a scene by making the desired object appear as white while everything else is black. In this case we wish to separate the city from the sky and therefore wish the whole city to appear white against a black background. The quickest way to do this is to under Matte Texture Element enable Material ID and set it to 0. All textures in 3ds max has their Material ID set to zero by default and hence every textured object in the scene (should be all objects) will appear white on the render.

Finally, check and double-check your settings and have that cup of coffee while waiting for the render to finish.

This is it for part 1. I’m a beginner at writing tutorials, but hope it wasn’t too long and painful. Anyway, this should keep your computer nice and busy for a while, depending on your system. If you want to take a peek at my finished file, you can download it below.

It’s been a hard week for the Green Spill team. Due to people shortage and some technical difficulties we hadn’t even started work on our previz when we arrived at school on Tuesday. As the previz is to be handed in tomorrow morning at nine, this left precious little time for animating, rendering, applying post and editing the thing. Add to that that most team members had other work during the few days we had, and you might understand why we were a tiny bit worried we wouldn’t finish on time.

We decided to give it a shot and quickly divided tasks between us: Jamie, Aleksander and Andreas would do the animation, producing a total of 67 shots. When they finished a shot, they would hand it off to me who would render it and take it through post. Lastly, Andreas and Alexander would do the editing and I would assemble the DVD needed for the deadline. To keep everyone sane, Andreas suggested that I’d make a list of all the shots in an Excel document where each person could add which shots they were working on and whether they were finished or not. By the end of Tuesday, the list was made, roughly 15 shots were finished, rendering was well under way and I had created a template in Nuke to speed up post-production.

The next three days were a blur where I’m concerned. I got up in the morning, started rendering incoming shots, did some post work, added more incoming shots for rendering, did even more post work and so on and so forth. I then went to bed, slept for an average of four hours, got up and repeated the process. I finished the last shot at 3 AM on Saturday. Three hours before I had to get up for work. During the night, I uploaded all the shots to our Dropbox so Alexander and Andreas could do the editing later that day.

And so we finished on time. The sounds and music are mostly temporary and some of the cuts aren’t quite tight enough yet, but it’s a draft 1 previz and we’ll get feedback on how to improve it tomorrow.  Today I’ve been assembling the DVD and designed a label for it for fun:

To create the label I had to do a high-res render. Thought I’d put it up here. Hope you like it Aleksander

Still reading? Then I’ll let you in on a tiny secret: The highest of our mountain peaks are 14 meters tall, only 8 times higher than our Luna. The render below is made from a camera angle that shows this quite clearly. My reason for using “miniatures” is to make the scene smaller and lighter to work with. To make them look taller and more distant I’m washing out the coulors a bit and applying some depth of field. Here’s hoping people won’t notice

Yes. I know, it is way, way too long since my last post. A new round with the flu and a death in the family slowed things down considerably for a while. I’m back now though and we have charged into the previz. A little late maybe as the deadline is on monday, but we’ve gotten a lot done the last couple of days and I believe we’ll be all right.

While Aleksander, Andreas and Jamie are all working hard, squeezing out shots, it is my job to render them, do some polishing in Nuke and have them ready for the edit on Friday (hoping we’ll be ready by Friday).

So in between receiving scenes and rendering them, I run the ones I have rendered through my Nuke pipeline. The process goes like this:

The 3D render is imported into Nuke.

As the render does not come with skies, I add those first. I found a nice image for this over at CGTextures. I add the sky separately, so that I can easily animate clouds, change sky colour etc.

To add the sky, I use a matte that comes with the 3D render. White determines what is visible, while black determines that which is invisible. By applying the matte to the render and positioning the sky image behind it, the sky gets added to the shot.

Next up is to add some depth to the scene using a depth map that also comes from the 3D render. A depth map determines the relative distance between different objects based on brightness values. Dark colours are far away while lighter colours are closer to the camera.

Based on the depth map, I am able to focus on one particular part of the scene. I then tell Nuke to keep this part in focus while gradually blurring the parts of the image that is further away. The result is depth of field, or DoF, that can be applied much faster than in 3D and can be adjusted after the 3D render. Definitely handy!

Finally, I add a sense of atmosphere by inverting the depth map so it looks like the one above, change the colour to light blue and add it to the render to wash out background colours. This is importaint to make the mountains look like they are further away. This is actually done before applying DoF, but needed to explain what a depth map is first.

When I’ve done everything above, I colour-correct the footage, blur it slightly, add some grain and render the result to Apple ProRes format, which we’ll use during the edit. The final Nuke script looks like this:

I repeat this process for each shot. For the most part I only need to replace the files from the 3D render and reposition the sky.

I haven’t had time to write anything the last few days unfortunately, so I’ll try putting the essentials down in this post.

Andreas had a few suggestions about the temple exterior, so I made some changes to it at school on wednesday and thursday. I then put them on my memory stick so I could work some on it during the weekend and left school. Without the memory stick. No update on that front in other words. I’ll put it up in its current version at my earliest convenience. Whenever that may be.

Not being able to work more on the temple, I tried fitting it into some surroundings (I’ll replace the temple with the current version when I find my memory stick). As this is still the concept phase, 3D can take too long to do so I ended up finding images on the internet and mashed them together after a fashion in the top image of this post. The shadows aren’t right, but not too important in this case. I’ve made the temple, the road and the stars. The other components I’ve found. Landscapes were made by Ladyofspira. I don’t have anyones leave to use the images, but as it is only intended as a possible guideline for general looks and meant for the team only, I hope I won’t get too many lawyers banging on my door. If I’ve used your imagery and you want it removed, leave a comment here and I’ll take care of it at once.

On thursday, the group made radical changes to the story, for the better in my opinion. Jamie wrote the second draft of the story while the rest of the team was present. Andreas and I worked the story into a draft one treatment that afternoon:

Synergy

FLY-OVER CAM. Futuristic motorcycle driving high speed across desert landscape, suspicious DEVICE with timer countdown fastened to bike. Bike approaches destination, lights from fire in distance. Bike slows down, stops in front of entrance to HIGH-TECH TEMPLE, signs of recent destruction and mayhem. FEMALE dismounts bike, picks up device and moves into temple.

Female moves deeper into temple ruins. Inside temple enters LARGE CHAMBER showing recent signs of havoc and destruction.
Giant PILLAR thrown towards her. Barely dodges flying pillar but drops device. MALE in armour appears in end of room. Female spots giant ROBOT emerging behind him. Robot backhands Male out of the way, charges against her.

Female runs against charging robot, leaps to attack. Attack dodged, giant FIST punches her, throwing her across room.

Male’s attention drawn to device. Runs towards it but stopped by robot attacking him. He attacks back, misses, robot throws him across room, through BRICK WALL. Female lunges for device, stopped by robot grabbing her. Robot pins her to floor with foot, about to crush her with fist, but knocked back by flying AXE hitting its face, from Male emerging from rubble. Female free, grabs SWORD and thrusts it through robots head.

Robot shuts down, falls to ground. Heroes stand over destroyed, inactive robot. Long awkward SILENCE. Countdown reaches zero.

Device gives distinct BEEP, breaks silence.

Female: (Neutral tone) Your lunch is ready…

Male: (neutral tone) Thanks. (Pause) So what’s for dinner tonight honey?

Female: (in same tone) Leftovers…

Male: What?! You said you where making meatloaf!

Female: (Irritated) Yeah! And you said you would fix the sink! That was last week! And what about the dog?! He smells like week old crap!

Couple walks away, arguing like typical married couple.

Finally, since the arrival of the aforementioned device, we’ve been in need of a count down display. As compositing etc. is my area, I spent some hours this weekend trying to create such a count down in Nuke. I had to do some light programming, something I haven’t done in a long time. Luckily I got some help from the good guys over at fxphd and I got it working in the end. This is a temporary design and will probably change, but decided to put it up anyway:

Tomorrow we’ll work more on the storyboard for the presentation coming up this coming Thursday.

My job during the weekend was to create a test for the opening credits/establishing shot which is a rip-off from The Day After Tomorrow. Normally, I would list what still needs to be done in the shot, but since this is such an early test, I’ll sum it up in one word: Everything! We’re not even sure we’ll use a shot like this and even if we do, none of what is shown above is likely to appear in the final shot in any shape or form. I made it because we wanted reference for the storyboard and to find a good technique  for creating a similar shot during production.

On a different note, the cinematic logo is more or less finished and can be watched underneath. There are still issues with the paint behavior, shot layout and music, but if we don’t get time to revisit it, this might just be the final version:

Tomorrow, Jamie will return from his holiday and we’ll kick the concept art part in to high gear to finish characters, environments, storyboards and animatic on time.

It’s been a busy couple of days and therefore little activity on the site. On monday we presented our film idea and got some feedback. Might work some of it into the film. We then shifted focus to concept art as we need to have characters, environments, storyboard and animatic ready by the 12th of February.

We worked from home on tuesday as Jamie was still in Egypt and Andreas had to work. I spent my time revisiting the cinematic logo, working more on the fluid sim and started a high quality render (first image). Included a Z-depth pass to bring the other paint buckets out of focus (second image). May remove the extra buckets all together.

I also went on the hunt for temple references. Found that I really liked the designs of Santiago Calatrava. Might try convince the other team members to use some of his buildings as reference.

Lastly, I looked at ways to create a desert landscape that we’ll need for the beginning of the film. Just a test at this point, but what I have so far looks like this:

Turned out the render was faster than expected, so stayed up and finished it. Really bad render quality, but I’ll rerender later. We had a 10 second limit on this – the animation is 10 seconds while the audio is longer. Hoping that’s enough of a loophole

Now I really must go to bed.

Due to various reasons, and against better judgement, I didn’t get around to continue work on the Green Spill cinematic logo before today, one day before the deadline. As this is my first real go at Realflow, I spent most of my day trying to solve the problem of why both brush and paint went right through the paint bucket walls, no matter what settings I used. I managed to find a solution in the end, but by then it was three hours to midnight and I needed to simulate the fluid, turn it into mesh, finish my scene setup and render the darn thing. I also need to run the result through Nuke to add a transition at the end. It’s now on my to-do-before-school-list, which isn’t such a great thing. This is why the fluid isn’t as “paint-like” as I planned. 

Rendering with the settings I wanted would take way too long, so I had to take out some background elements and render with crappy settings instead. Will simulate and render out a better version later to put in front of our upcoming film.

I’ll post the movie here when the render is complete. That’ll be in the morning. Now I better go to bed if I need to get up at five…

Rig: Brad Nobles Skeleton Rig

The second assignment is about facial animation and lip sync. We were to pick a sound file and animate a character based on it.

Rig: Enrique Gato Borregáns Tito

Bear in mind that these clips aren’t necessarily the finished ones and that character animation isn’t what you would call my cup of tea There’s also a third digital acting assignment that I’m working on. Will post it when it’s closer to a finished product.

“The Diversion” is an animated short film about an energetic old woman, Rose, who wholeheartedly wishes to enter a night club. When the guard at the door refuses her entry, she devises a series of clever schemes to get to the door without the guard knowing…

So the movie is finally done and we’re reasonably happy with it. Some of the animation, voice acting and rendering could definitely have been better, but there is only so much you can do in four weeks. I’ll update this post with more information when I’m less exhausted

The teacher told us to create a simple 3d character for use in the upcoming animation lessons. I came up with this one. Made a face for him too, but that’s not quite ready yet. I prefer to keep him anonymous for now. Rendered with V-Ray and and lit with an HDR image. The resolution of the HDR isn’t that great. Hence the grain.

Ok, project’s over and site’s updated so I thought I’d do a write-up as promised in an earlier post. Those who know how to count might have guessed that this was the third big assignment given at 3D design and animation at Noroff 07/08. As such, we were once again to build on skills already acquired and character creation/animation were the natural progression.

Criteria for this project:

I also made a carrot as a prop, to emphasize the fact that the character is a vegetarian. Although it’s almost too small to see in the final shots, I thought it turned out rather well:

I had a fair bit of trouble when it came to do the hair. I experimented with everything from mesh to Hair&Fur to VrayFur. All of which had their own advantages and disadvantages. In the end I stumbled upon a plugin called HairTrix which seemed to play fairly nicely with V-Ray:

A bit more detail modeling and texturing and the character was ready for rigging and skinning:

Due to too much experimentation in the beginning of the project, the fact that one of our four weeks was the easter holidays and because my first skinning attempt disappeared under mysterious circumstances, there was now very little time left. With three days to deadline there were still skinning for a second time, walk cycle, poses, rendering, DVD and presentation left to do. To say I was rushing things near the end is a little bit of an understatement. Still I finished on time after working through the last couple of nights.

Pose 1:

Pose 2:

In this pose the character is supposed to be inches from touchdown. It looks a bit weird though, as I didn’t have the time to create the chute itself.

Pose 3:

This was the last pose I did. I wanted the character to look skywards, but not entirely sure I pulled it off. Made only a few hours from deadline.

DVD:

Walk Cycle:

Due to some miscommunication between me and my teacher I got the impression that one cycle (every two steps) had to be made from 24 frames. This resulted in a rather faster walk than I would have liked. When I finally realized we could choose the cycle length ourselves, there wasn’t enough time left to fix the cycle and re-render. The character actually blinks in this cycle, but due to the size of the movie it’s rather hard to tell.

All in all, I’m quite happy with the result although in hindsight, there are certain things I wish I had done differently. If I had more time, these are the first things I would have improved on in no particular order:

I’ve added a few more points to the list above. Got my grade this week as well. Another A

Finished at last. Well, nearly. Still need to come up with three still image poses + wires, design the DVD cover/label and create a keynote presentation. Good thing the deadline is still weeks days hours away. I’ll do a more thorough write-up when on the other side of the deadline.