Hello all,
I am looking for resources (tutorials, articles, research papers, books, etc.) about ART QA (not necessarily game QA). Basically QA practices and tools which aid the art creation process. I’m not really looking for game QA testing, where the game is tested as a whole, but the steps before.
I’d be interested to hear how QA is handled during different production steps (modeling, animation, etc). I’m interested in tools (e.g. check tools), processes (approval systems, checklists, etc).
I’d also be happy to hear about production horror stories of how reworking assets to death could have been prevented with some oversight
Why my interest? I would like to research current art production QA practices for a research degree, and come up with ways for improvement, and possibly try them out at my company in a production setting.
Thank you!
edit: I would also be interested in wider quality topics, as long as they relate somehow to art production, such as lean, kaizen or other management approaches and philosophies that have the aim of promoting quality and customer satisfaction.
I haven’t used it personally, but Pyblish looks pretty neat. It’s a framework for validating content.
At SOE, on Star Wars Galaxies, we had a Technical Art Director who personally reviewed every art asset and approved the poly count, texture resolution, shader complexity, etc. There was also a 7 page checklist. It was kinda crazy, but necessary since we were maintaining a live service and an art asset could cause massive performance issues.
At a previous job, we had a pre-export validation step that would prevent exporting of assets unless they passed all the tests. It arguably caused more problems than it solved, since it slowed down the iteration process. Of course the artists hated it at first, and I like to think they got accustomed to the common errors and learned to work smarter to avoid them.
I wouldn’t call it a ‘resource’ , but here’s a tip
you want to have a unified URL system for referring to things – one that supports cut-and-paste and works in links. It’s especially important that the URL be included in things like debug screenshots: for example, on Class3 I had a little tool that took a snapshot , recorded the camera’s position, and included the URLS for all the assets within a certain distance of the camera then uploaded them all to a bug database. That way you could report a bug on an asset (a misaligned texture, bad shader, whatever) without having to hunt around in the 10,000 asset database to find out what was. The person triaging the bugs could click on the link in the bug report and see the history of the assets and thus route the bug to the appropriate artists. In more manual systems the time cost for both the reporter and the responder to find the actual problem asset can be quite large – verbal descriptions are muddy and no company does a good job of naming and organizing folders
Thank you for the answer, so far! I’m really trying to get a holistic picture here about practice, theory, software tools and process
Does anyone of you use mathematical or statistical models to analyze your production metrics regarding QA, in order to track down areas where the production can be improved? E.g. how often an asset needs to be re-worked; how often assets are approved by mistake; how often assets are bounced back from the following production stage; how much time is used for the QA and approval effort itsel? - with the goal to track if your production and your work’s quality is getting worse or better.
Does your studio have some sort of self-improvement plan regarding quality that is being followed during production? (i.e. not a post mortem, which happens after production is over). I’m thinking of the lines of “continuous improvement” methods, like CMMI or similar, or TQM approaches?