Install Steam
login
|
language
简体中文 (Simplified Chinese)
繁體中文 (Traditional Chinese)
日本語 (Japanese)
한국어 (Korean)
ไทย (Thai)
Български (Bulgarian)
Čeština (Czech)
Dansk (Danish)
Deutsch (German)
Español - España (Spanish - Spain)
Español - Latinoamérica (Spanish - Latin America)
Ελληνικά (Greek)
Français (French)
Italiano (Italian)
Bahasa Indonesia (Indonesian)
Magyar (Hungarian)
Nederlands (Dutch)
Norsk (Norwegian)
Polski (Polish)
Português (Portuguese - Portugal)
Português - Brasil (Portuguese - Brazil)
Română (Romanian)
Русский (Russian)
Suomi (Finnish)
Svenska (Swedish)
Türkçe (Turkish)
Tiếng Việt (Vietnamese)
Українська (Ukrainian)
Report a translation problem
SFM has next to no multithreading support (I think it uses a second thread for some things, but only some), but is very CPU dependent (due to shadow mapping), so while your CPU is about as good for SFM as is possible, you're going to be seeing no benefit from the multiple cores or the high-end graphics card.
Without knowing more about your export parameters, such as frame rate, resolution and sample rates, that doesn't sound wildly unexpected.
You should bear in mind though that this is still *staggeringly* fast compared to some other animation software. (On my machine, SFM can comfortably render even intensive frames in a few seconds. Something like Blender or Daz Studio will take several minutes per frame).
Res: 480 (cos it's a test)
FPS: 60
I don't know my sample rate, sorry! I guess it's just a case of waiting :)
Combine that with not having to render everything in one go* and being able to re-render edited parts (also without having to re-render everything else), and you can save a lot of time with the export process if you have to stop the export part of the way through and/or want to change something.
* I still recommend exporting from the start of a shot to the end of it (I personally recommend using the "Selected Shots" export duration) rather than starting/stopping halfway through a shot, but it's completely okay to start from the beginning of a shot, even if that's halfway through the session.
But as Marco Skoll said, uninstall QuickTime as soon as you are able to. It's officially considered a threat for Windows,[www.us-cert.gov] and it's not really useful for much besides Source FilmMaker, and even then it's not even useful for that anyway (due to how poorly Source FilmMaker exports MP4/MOV videos, especially compared to image sequence exports).
... well, entirely dependent on your sample rate, an hour to render that properly doesn't sound wildly off.
For example, this scene is particularly intensive (rendered at 720p60, 64 samples, with very heavy use of shadowed lighting):
https://www.youtube.com/watch?v=Ov87eDJLaDM
(It has since been royally buggered by Youtube's compression - 3000kBit/s is just not enough for swirly camera moves and loads of particle effects).
I originally rendered that on my old computer, although I've since re-rendered on my current machine (which, while not quite as strong as yours, is still very solid for SFM - the i7 4790 is still a pretty formidable CPU for single core performance, and the GTX 1050 Ti is definitely not holding it up*).
Those 8 seconds take 48 minutes to render.
* I can say that absolutely positively, because I've used this scene as a test on several occasions. On my old machine, there was less than a 2% difference in render times between an HD 7570 and a GTX 750 Ti. (Which is a considerable upgrade - it would double the frame rate in many games). This scene is almost entirely CPU limited.
The AVI function in SFM is a broken tool and does not work most of the time.
Export an Image sequence Movie from SFM and use a 3rd party software to stitch into a proper AVI or MP4. This guide shows how to do that using Blender as a Video editor. http://steamcommunity.com/sharedfiles/filedetails/?id=375229570&searchtext=blender
Also, exporting Image sequences produces much better results than exporting any type of "movie format" direct from SFM.
First thing you need to do, as Marco pointed out, is get rid of QuickTime for Windows!
It IS a legitimate security risk, and if anyone tells you that it's still ok to use, they either want you to get hacked or are taking advantage of your lack of knowledge to hack you themselves...
Have a look: https://www.us-cert.gov/ncas/alerts/TA16-105A
Use image sequence render, it's much more reliable as far as quality control is concerned.
Another thing that will increase the render time, regardless of method used, is Framerate and Resolution. I like to render in 1080 HD at 60 frames per second (fps) but the render time, even for a short video, at those settings is huge!
I dropped back to 30 fps and that did speed things up noticeably, but it still takes a while.
Best advice that I can give is give yourself plenty of time, not only to do the animating, scenebuilding, etc., but also for the render as well. This is not something that you can just toss something together with in a couple of minutes and expect to win an Oscar.., lol, it just ain't gonna happen that way, and the video pages are littered with videos where that's what was done, a rush job. Yes, I'm even guilty of it, and I have regretted it ever since... lol
Aside from a short moment of normal HWM male Scout, then yeah it is all that's there, aside from obviously some sound clips.
Yeah it's at 512. I had it on the default before and whilst it did go quicker I was getting grainy renders that just looked horrible.
Gotcha. Will try that, thanks!
While I generally render still images at 1024 samples (at 4K), that's just a case that when I've been posing and lighting a single frame for hours, it seems stupid to worry about a two minute render time.
Generally, unless a scene has extremely shallow depth of field, you're unlikely to notice more than 64 samples.
~~~~~
Ramble time!
I did a comparison once, comparing 32 samples, 64 samples and 2048 samples*
* I rendered at 1024 samples twice, then merged them. Because the sampling is random, the two images were slightly different (provably so, if compared in Photoshop), and so the result is broadly equivalent to if SFM were able to render another quality level higher.
http://steamcommunity.com/sharedfiles/filedetails/?id=831086133
http://steamcommunity.com/sharedfiles/filedetails/?id=831087144
http://steamcommunity.com/sharedfiles/filedetails/?id=831087697
I can just about see a few differences between the 32 sample image and the 64 sample image - slight smoothing of the railings in the background, as well as slightly better ambient occlusion.
However, even toggling back and forth between them, about the only difference I can genuinely see between the 64 and 2048 sample images is a slight difference in the ambient occlusion on her neck that I definitely wouldn't notice when it was zipping past at 60 fps.
Honestly, if you want a setting to turn up high, turn up SFM's anisotropic filtering; the default setting is pitifully low given how fast modern graphics cards can make mincemeat of it (and this is GPU load, which is utterly negligible to how fast SFM runs anyway, particularly if that GPU is a 1080 Ti).
This is what the scene would look like with SFM's default anisotropy settings - a noticeable fuzziness to oblique textures like the railings and brickwork in the background:
http://steamcommunity.com/sharedfiles/filedetails/?id=831088778
Oh wow, honestly even on a 4K (27") monitor, I'm not personally noticing any difference between any of those images, even if I get right up close. I might knock it down to 32 then, just to see if it makes any difference on video.
I test rendered (on 64) about 15 seconds of footage, in 480p, and the video was over 2GB which is borderline insane lol. My internet can handle a big upload to YT but whatever makes things 'smaller and faster', as it were, is a plus for me! I'm going for the Extended category in the Saxxy's, so 5 minutes at 720p+ is going to melt my internet connect lmao
I have my videos at 1280x720 (though at 30 frames per second, not 60) and have about 50 megabytes per minute in a video (I think)... and even then I'm aware I'm using relatively-high quality settings considering that I intend my videos for YouTube (which compresses uploaded videos a lot or something) and that my videos are not super fancy.
You can see some examples of different filesizes/compressions compared to the end product in this video, I think: https://youtu.be/hRIesyNuxkg (It's a YouTube video, but as it's best viewed in a browser outside Steam forums, I've given a normal link to it instead of embedding the video in this comment.) Granted, that does go into a lot of more technical fidelities than you'll probably have to care about, I think, but it's worth a watch in my opinion.
That sounds like an uncompressed AVI, which will cause you BIG problems.
Once exported as an image sequence and compressed into MP4, the full 8 second scene here (in 720p60, so broadly equivalent to 18 seconds of 480p60) was only 50 MB, even compressed at higher than Blu-Ray quality*
*Like I said before, the version I linked before was compressed to buggery by YouTube, but the quality in my master video file is all but indistinguishable from the 64 sample images I've linked above, other than some colour correction in post - this is the same frame extracted from my master video:
http://steamcommunity.com/sharedfiles/filedetails/?id=765081871
(As you may have determined, this scene of mine tends to be something of a benchmark for me to test quality settings).
Actually, the file size will be (in a compressed format, anyway) slightly smaller for a higher number of samples, as it reduces the grain in the image, and thus makes it slightly easier to compress.
We're not talking a massive difference (just a few percent, even for quite large increases in sample rate), but it is there.