Friday, June 15, 2012
Visual Effects Tips 2
[Visual Effects Tips 1 post]
Time is important in visual effects. There’s never enough of it so make every attempt to minimize wasted time and simple mistakes.
Many of these examples are animation or compositing but most apply to the majority of visual effects work from texture or matte painting to lighting.
When starting a task or shot check the materials you’ve been provided.
Are all the materials correct?
Does the live action match the count sheet?
If there’s a match move is it correct?
If you’re composting are the supplied renders correct with correct alphas?
If there are problems flag them immediately. If you don’t check you may not know that there are problems until well into the shot, at which time trying to correct the mistakes will take longer and be more difficult.
After confirming everything is ok become familiar with the material. If it’s a greescreen take a look at the screen and potential problems extracting the mattes. If you’re going to roto the live action run the clip a few times to see what’s happening so you can plan your approach.
Note to visual effects companies: Make sure your contracts include explicit details for deliveries to your company. File formats, schedules, naming, file organization, size, etc. Sometimes it’s amazing the amount of incorrect material received from a client. On one project I worked on the editorial dept seemed to have a PA who was simply recopying everything onto hard drives and shipping them with no information and no organization. The visual effects company then would have to spend hours sifting through the material to determine what was new and what had changed. In other cases I’ve heard of some editorial departments constantly providing incorrect count sheets. By having this info in the contract the first time this happens flag it to the client and inform them that additional lapses that require rework or unplanned man hours correcting will be billed on a time and material basis. Additional costs seem to be one of the few ways to get a clients attention unfortunately but it also means that the meager profits will be less likely soaked up by these types of sloppiness or errors on the clients part.
Now that you’ve reviewed the materials take a moment to think through the task at hand and how this will be accomplished. What are the critical elements, what are the steps that will need to be done, what tasks will have to be done by other departments, etc. What are the items that will likely change or that will need special care? If this a stereo shot or will be converted to stereo after the fact then you will have to consider the implications of working in real 3D space. If you’re compositing you’ll need to think of the layering process.
Keep it simple stupid. In your plan for the task try to keep it simple. You have to anticipate the areas that will need to be controlled independently and break it up accordingly but avoid going crazy with layers or keyframes if possible. A complex shot will probably require more than simple work but don’t needlessly make the project or the work complex.
As you start thinking of the shot and the steps required keep in mind it’s likely you will have to pass the work on to someone during the course of the show. That someone could be you having to re-open the project and make changes two months after it was finalled or it could be you re-starting the shot after it was put on hold for three months. Or it could be a co-worker who has to take over the shot while you finish something else or the shot may have to be split up because of an impending deadline.
With that in mind you want to make it so no one, including yourself, has to spend hours reverse engineering what you were doing and why. (I’m not even going mention the possibility that a sequel may mean some of these files and projects are opened years later)
Start with the fundamentals. If you’re an animator then you want to make sure you have the proper poses at the correct key frames. Don’t get lost in trying to do the secondary animations, just focus on the fundamental animation. If you’re a compositor adjust the black levels and adjust the basic color of the assortment of elements you’ve been provided before throwing in all the layering atmospherics and other enhancements.
Once these are done and approved then you know you have a solid foundation to build on. It allows the supervisor or director the ability to review the work before you spend a lot of time on it. There may be radical changes that would make all the additional work worthless.
There will be times when it will be useful to do slop or rough animation, renders and composites. Frequently it’s necessary to block in an entire sequence so the director and editor can review before proceeding. Other times it may be necessary to provide shots for test screenings that were scheduled to be done in the future. Regardless you’ll get notes on the shots and the first impulse will be to take these quick and dirty versions and start making adjustments. However it’s almost always best to get back to the foundation and the basics before addressing the notes. If the black levels are off or the basic timing of animation is wrong, it will still be wrong even tweaking the other aspects of the shot. And in the end it’s likely you’ll waste a lot of time and have to scrap it to get back to where you should have started from.
You should have a To Do list for every shot you work on. With feedback from the supervisor or director you may be changing the order of priorities. Neither the director nor supervisor will want to ask multiple times for the same thing. Let them know if there is an issue accomplishing that.
If you’re reviewing yourself play the shot through and note the items that stand out to you as not in keeping with what the final shot should be. Determine the priority based on which issues truly jump out the most and which ones are fundamental issues versus getting sidetracked and spending time noodling with a few pixels in the corner. If it had to go in the movie tomorrow what are the top 5 items to be fixed or changed?
As you work on each item mark it off your To Do list so you always know how much you’ve done and how much is still to be done.
Is your monitor actually calibrated? Are you using the correct lookup table and gamma settings? Are you seeing the same results you see in dailies review and that the supervisor and director see? If not check with your company to see if your system can be calibrated or you’ll have to develop an eye to compensate for the difference. Be aware if your system or playback has limited color depth.
If you’re presenting to the director or supervisor it’s usually worthwhile having the previous version loaded and ready to review since many times it will be used for reference and comparison. If you’re showing with and without results in something like Nuke then wire up the viewer so you can quickly cycle though the different versions or variations if possible. Having to zoom out, find the areas, relink the viewer, etc are time consuming and should have already been done ahead of time. Sometimes it’s worth putting in a switch to quickly change the composite flow. Create snapshots in After Effects or other apps.
Reviewing your work
For your own reviews crank the gamma up and down and the exposure up and down to spot any inconsistencies or level issues. Try the alternate lookup tables (make sure to return to normal once done reviewing). Will the shot hold up in the DI or once it’s transferred to video? Are areas being clipped or is there a mismatch of blacks and levels? Is there any quantizing? Check the blue channel for grain issues. Have you added the proper amount of grain and does the whole shot match?
Play it slow, fast and in reverse to see if everything is working as it should. This can sometimes show flaws in the animation or simply allow you to spot fluctuation problems.
When you’re staring at something a long time you may not spot some of the flaws that might be obvious to others. Painters sometimes look at their work in a mirror to give them a different feel for the painting. This was common in the pre-digital days with matte painters. Flopping the image in the viewer or composite will provide this same option of seeing the shot in a new light.
Be honest with yourself. If the animation or whatever you're working on isn't working consider how to fix it. Discuss with a co-worker, lead or supervisor if you know it's wrong but aren't sure how to proceed.
Do your work in an organized fashion that follows logical thinking. If your company has templates or guidelines be sure to follow those. It’s easy in some apps such as Nuke to end up with something that looks like a mess of spaghetti. This will chew up your time daily to maintain and change and it will mean that anyone else who needs to work on the composite will spend extra time digging through the composite.
Come up with some basic conventions yourself if none is provided by your company or lead. Try to be consistent about the approach and naming. Group parts of the shots as modules. This makes it easy to enable/disable sections or to switch sections. If you’re working in Photoshop you might want to put sections in folders to keep it more organized. Likewise with some roto packages and other apps there are opportunities to group related items.
It’s not always clear what your project file is doing. It may seem totally obvious to you now but to others or even yourself a few weeks from now it will likely be a totally mystery. If you’re writing software code then include comments. If it;s a complex process with multiple files consider writing a Read Me file or some basic document describing what the process is and how it works. By including some forms of documentation it will be easier and faster to keep track of what is what. Many software packages offer methods to document the work as you’re doing it. Label layers in After Effects and Photoshop. This minimizing the amount of guess that is done or the toggling of layers to determine exactly what each is doing. Label key nodes in Nuke.
Some apps allow adding comments. Nuke provides Sticky Notes so you can mark a specific area with a reminder to yourself or an explanation why you’re doing it this way. Their backdrop can be placed behind a number of nodes with text and with user defined colors making it easy to spot key areas even on a large and complex composite. Nuke also provides No Op nodes that can be labeled. Take a look at the software you’re using to determine how you can use labels and other built in documentation. If none exists consider writing up a short text file in the director with your notes.
Add slates to your rendered images and add useful comments there. These are going to become very valuable when you have to dig back through them for yourself or a client.
Many software packages have autosaving functions. Make sure these are turned on but also get in the habit of saving yourself to make sure your work is being saved where it needs to be saved.
Save your files and projects as you go along with version or take numbers. Use the company defined file structure and naming conventions. Especially important if you’ve made major changes. It’s very frustrating when the client or supervisor likes a test or previous version you did but you didn’t save the project. Now you have to spend time trying to recreate what you already did once before.
When you work on a shot you will likely have to render quite a few takes on your machine and/or on the render farm. If for some reason your render isn’t optimized that can equate to quite a few hours of lost production time and processor time. As you manipulate, test and review interactively an unoptimized configuration can mean the difference of trying multiple changes in a short time or become a slog of spending most of your day making a few simple changes. It may mean the difference of producing a great result on a short turnaround or a mediocre result since you didn’t have time to manipulate and render a better version.
Optimize your renders when possible. Know the application you’re working in and know what options there are when rendering. Here’s a list of optimizations for Nuke. There’s likely more than a few lists for all visual effects software packages. Avoiding things that slow down the app and doing more of the things that speed up the app may have a big impact on the day to day work. Consider using low resolution proxy models or elements when you can for working on animation or basic testing. Consider pre comps and caching when working on the top layer. Consider rendering just a frame range of the shot and combining it previously rendered frames to check the new frames in context.
For specific tasks you might turn off some of the advanced settings - motion blur, fur render, etc. as long as you know what you’re evaluating. You might be able to disable specific areas of a composite or render if that’s not what you’re working on. Just make sure to restore all of the settings for the final render and another other renders that require full reviews.
There’s no point in spending 3 days trying to shave a few seconds off the entire shot render but if your render is running much slower than it should due to simple changes, then it’s a waste of time and resources.
The other thing to consider optimizing is yourself and your approach. If you’re working a lot in one software package then get to know all the keyboard and mouse shortcuts. Configure the app keys or your Wacom to take advantage of specific and frequent tasks. Use macros or scripts as needed for repeating common tasks. In Photoshop use actions for tasks that you do a lot of.
Check your results
Check all the renders or tests you generate and consider them in terms of the request from the director or supervisor. Make sure to check your work before passing it on to others to use. Do NOT assume that its correct. If you’re an animator check the final animation for any stray keyframes or oddities. Rotoscopers should make sure the motion and coverage is correct. If you’re rendering simulations make sure they’re the required length and have correct alphas. If you’re a TD make sure there aren’t any glitches or bad frames rendered.
It’s far better to find out now and fix it rather than getting a frantic call late at night two weeks later.
Once you’ve checked the results make sure everything has been correctly entered int whatever asset database your company may be using. Make sure to prep the files for archiving if you’re finished.
Keep a log
Keep a log of the time you work and the shots and tasks you’re working on. You likely already do this for your timecard or for production but it’s worthwhile recording it as well for yourself. It can be a real eye opener to see how much actual time was spent on a shot or given task or how much linear time it took by the time the back and forth happened with the director. Use this as a reference when you’re asked to bid on a shot or when you’re asked how much time you think is required to get the version done. All of us in visual effects are eternal optimists. We assume we can whip out the next version in a couple of hours. Time after time of saying that with a client waiting and disappointing them with something two days later is a problem. It’s also a problem for the visual effects company who bids on the amount of time required and unknowingly underbids the work simply because everyone is optimistic. The company should in fact be doing their own reality check of what the typical shot requires but most companies don’t work that way. No point in budgeting 2 hours of roto per shot on a new show if the company average was 8 hours of roto per shot on the last show.
Also keep a record of the bids you provide when bidding a show. Don’t be surprised if the numbers changed quite a bit when you go to do the actual work months later. Especially problematic when they try holding you to it. The supervisor or producer may think you were pessimistic and ‘adjusted’ the amount of time much lower. Or in some cases they may have added a lot of padding.
Check the next phase
Once you’re done with your tasks and passed the files or elements on it’s usually out of your hands. And the company probably wants you to be focusing your efforts on the next task. But when possible it’s worth checking the next step, the next render or task. You can check to make sure the correct version is in fact being used of your animation. You can make sure the root elements or renders are the right ones and are being used correctly. It’s possible there’s been a mixup or confusion on the next step and by viewing the 10 second shot you can flag an error that may not be evident to others.
Check the final
If a shot you worked on in any form is finalled check it out. Its worth doing a reality check of what you did and how it was used in context of the shot and project. Did you over build details of the model that never show up given the action or smoke levels? Did texture map you painted work? Did that elaborate 30 second shot get cut down to 2 seconds? Would you have done more on a specific aspect that would have made the shot better? There may not be much you can do about some of these things but it’s worth getting the experience and understanding how things come to be. It’s also worth considering how would you approach this same task or type of shot next time. Would you change anything? Have you learned anything you can apply to the remaining shots or to future projects?
[I'll have future tips posts. Feel free to add your own tips in comments]