Animation and background movies
thoromyr
Posts: 452
So... Daz doesn't care that using a movie as a background is broken. Oh well, in my case it just saves a step (making the movie) so using an image sequence (thanks to Wendy for the tip!) works great. Except...
After some testing I can't come up with a general set of steps to reproduce the problem, but the short of it is that Carrara stops treating it as an image sequence and uses only the first frame. As noted, I haven't been able to get a reproducible set of steps for the problem, but it appears to be set in stone for the scene I'm working on. I was, temporarily, able to get it to be an issue for a new test scene after switching image sequences a few times, but closing and re-opening it works just fine.
Things that *seem* to impact image sequence becoming image still:
- set sequence for background, render test, switch, repeat. Around fifth it stops working
- set large image (8000x4000) sequence as background
Symptoms:
- Carrara does not prompt for the FPS value
- Using the slider to move through the image sequence in Scene/Background has no effect
In summary:
So far I can use any image sequence with my test scene and it works, regardless of resolution. I have only been able to reproduce failure in test scene by switching between different image sequences, but restarting Carrara restores functionality. However, the scene that I need this to work never works any longer, no matter restarts, saving without background, closing, restarting, adding, etc.
What I'm hoping for is that someone else has seen this issue and can shed some light on it.
Daz 3D is part of 
Connect
DAZ Productions, Inc.
7533 S Center View Ct #4664
West Jordan, UT 84084
Licensing Agreement | Terms of Service | Privacy Policy | EULA
© 2024 Daz Productions Inc. All Rights Reserved.
Comments
I can certainly see the reason behind wanting such things: An animated spherical 'background' rather than a backdrop. But try to understand that this is asking for some computer taxing stuff going on, especially if everything is enabled, such as reflections, refraction, etc., as spherical background images tend to use a minimum resolution of 8,000 by 4,000, which can still be too low resolution to look straight down or up. It's not as simple as just having an image anymore.
Part of this taxing affect can be noticed by loading in a single 12,000 x 6,000 image and watch how long it can take to load - but once in place, it gives a great deal of power to your surroundings.
Even when not using GI and/or IL, a background image will assist in lighting your scene, I think.
Anyways, I'm not entirely sure that this stuff doesn't work - but might just be a bear on your computer.
The more efficient method would be to plan out your camera movements ahead of time and render, or otherwise create, Backdrop animated sequences or full-frame movie files. You can then use a background still which will affect your reflections and such.
I could be wrong on this. But I began some experiments on just this sort of thing, and my eight cores and 16GB RAM were taxed to the max trying to help me achieve such a cheat, which is truly what we're trying to achieve when doing this sort of thing. But I'm not sure thatit's fair to blame DAZ 3D for not anticipating that this should be made easier on our resources, since doing these things are resource demanding because of "us", not Carrara, and not DAZ 3D. Just saying... and certainly not trying to sound or be argumentative. It's just a hefty thing to ask.
Mine save fine
Are you saving uncompressed using local settings?
It saves however I set it up in the timeline oscillated reversed etc
If I move the image file Carrara will ask for it loading the scene too.
They do not always update in preview though but render correctly
thanks for the feedback, both of you, I appreciate it.
Dart: You seem to overestimate the system load and under estimate my system. Its older, but I have 32GB RAM and this isn't actually (much of) a load on the system. If I take a 12,000x6,000 background image sequence and tile it 2x2 my test scene barely takes more than 2GB when rendering, and that's a peak -- it is nearly always below 2GB (just did this to test). A heftier scene (the test scene is *very* simple) would require more, but it isn't the background image.
Perhaps you were thinking about movies where the memory consumption goes up geometrically with the frame count? (And Daz keeps a poker face and insists nothing is wrong. Given the impracticality of using either movie import or export they might as well remove the "feature" -- it would be more honest).
As to rendering to backdrop only: depending on what is actually going wrong, that might not help at all. And wouldn't work for what I'm doing anyway. As I'm rendering the initial background in Carrara, at first blush I should do it all together and avoid this whole mess of backgrounds. The problem is that I'm manipulating the background after render and before use to achieve something not possible staying inside of Carrara. Well, at least not possible for something with my abilities... I'll freely admit there are people who can *really* make Carrara sing and dance (Tim Payne, for one), and I'm not one of them. To see what I'm doing check out this link.
Wendy: uncompressed? yes. And I take it you don't have an issue so something going on here locally that I still need to figure out.
Unfortunately I'm not sure when it stopped working for the scene I care about. Where it really stands out is the end of the animation when the camera is motionless and there should still be movement from the background. My first thought was wrong, so I wasted time chasing down the wrong rabbit hole.
And even my tentative "larger is worse" is wrong: as mentioned above I can take the 12,000x6,000 and it works just fine. Now I'm thinking it may be a different "larger is worse", the number of frames in the background. All of the ones that work are <1000 frames. The one that doesn't work for any scene is >1000 frames.
Of course, there is the scene where *nothing* works (any longer). Sigh. Although I have a fair amount of network storage, most of it is taken up making this a disk-less household (children are death on disks, all dvds are ripped to avoid damage and facilitate playback) and my scene files as a whole are too large to not exclude the directory from time machine. I already tried rebuilding the scene once (move assets to the browser, then into a new scene file) without improvement. I may have to start from scratch and do the whole thing over again. Sigh.
One other thing: the cseqm file -- AFAICT it isn't required, but if present it appears to affect the result of importing an image sequence. Anyone else know if this is right or wrong?
Well, I wasn't actually talking about the image "map", itself, being the resource clog. It's when you build into the scene.
Jeremy Birn does a beautiful job of explaining how "Ray-Tracing" is accomplished. While I cannot do the subject justice without more time to consider my words, the engine has to calculate each and every point of incoming light against what is in the scene to be rendered. At that time it must also calculate the angle to which the light intersects things, and how those things are set up to handle the light once interaction is made - sometimes the shader asks something else from the engine, depending upon the angle of that incoming light.
A simple three-point light rig therefore renders fairly quickly. I took some really genius programming to make that happen though. Because a lot is happening between the relationship of light, shader, and render settings. That's why I'm always going on about how much I love how "FAST" Carrara renders. It really does calculate ray traces very quickly, considering that its default ray depth is eight passes.
Anyways, the atmosphere tab affects ray tracing. This includes Background Maps. That's where I was going with my babble.
So, perhaps I'm not understanding your issue. I've conducted a quick test, and uploaded the result to YouTube: AVI in Background Test
As the video states: This AVI is meant to be in the Backdrop settings, not the Background, so it's quite out of scale. But it is an Full Frames avi, which is how I always save avi files that I need to process further. I never compress them to a codec until the final production phase, which is why I've taken it upon myself to start a Hard Drive collection! :)
Not to split hairs here, but unless they changed something in C8.5, Atmosphere and Background are under the scene's Effects tab. Background is more than likely raytraced on some level, but I think the light interactions are more complex with the atmospheres due to the different variables in the RSE and how scene lights interact with them. The Atmosphere and Background options shouldn't be conflated in my opinion.
Basically, the Background is calculated in the final render, as it can be reflected, although it doesn't act as a light source unless the GI component, Skylight is enabled. Plus the background does not react to other lights. I suspect with the standard renderer that the brightness of it is treated kind of like a glow. I don't know about movies or image sequences, but I can, with a large degree of certainty, say that a background renders faster than an atmosphere. Of course the backdrop render so fast because there is no light interaction or reflection.
Right. Rendering the background or backdrop doesn't even need to glow, as it's just transferred behind everything else in the scene.
...and, Right. It would be GI and IL that would effect the ray tracing - not otherwise... you're probably right... I stand humbly corrected :)
Rendering off a HUGE spherical image of a detailed 3d scene can take days for a single frame - but if we want to look up or down to any great degree, we need to either block the spherical image, or make it really, really huge, pixel-wise - turning an otherwise simple render into a very complex process.
So I did it anyways, without getting too huge. It takes a bit to load the image into the background. So I have a feeling that flipping through an animated booklet of those things on the fly will eventually clog up the bandwidth of something.
Well, I provided actual numbers to back up my statements as to system load instead of vague handwaving. I suppose I could've pointed out that the render times for each frame was <1 second, and that for the larger scenes (with reflections!) were definitely < 10 seconds. This is not a resource hog, even at high resolution maps (12000x6000 is larger than your postulated 8000x4000).</p>
As I said in my last post, it is apparent from Wendy's statement that the problem appears to be local. I haven't chased it down, but I got the feedback that I needed initially. If there were a known issue then I would know I don't need to investigate further.
ah. Since I knew backdrops wouldn't do what I needed I hadn't thought about that, but it is an excellent point. As can be seen in the actual scene I linked the reflections are an important part of the result.
To be clear: I'm *not* using backgrounds for speed: I'm rendering images, applying post processing and then using them as backgrounds. And, because I'm having to render with the spherical camera, I have to render at higher resolution than would otherwise be necessary -- that is, if I could do the whole thing in one pass.
Well, there's obviously some kind of raytracing even with the standard rendeerr I think. You need raytracing for reflections right? I just think it's a really simplified raytracing method. I just wanted to make clear to less experienced users that come across this thread that the atmosphere and background are not the same, and are more than likely not calculated the same at render time.
I should have clarified the glow thing a bit better. Maybe luminosity would have been better. What I mean, is that if you make a billboard or splat and put an image on it, but don't light it, it will be dark. If you copy the image to the glow channel, then you don't need to light it. The image in the background is fully visible without the need for lights similar to a glow, so I was speculating that Carrara treated it in a loose sense like a glow. Heck, with Indirect Light, an image in a glow channel can be an area IBL.
ah. Since I knew backdrops wouldn't do what I needed I hadn't thought about that, but it is an excellent point. As can be seen in the actual scene I linked the reflections are an important part of the result.
To be clear: I'm *not* using backgrounds for speed: I'm rendering images, applying post processing and then using them as backgrounds. And, because I'm having to render with the spherical camera, I have to render at higher resolution than would otherwise be necessary -- that is, if I could do the whole thing in one pass.
I kind of figured you were doing it for more than just speed. ;-)
If I may offer a possible suggestion, you could try and lower the resolution of the image and apply it to a scene encompassing sphere. Use the spherical image sequence in the sphere's color and glow channel, and limit the light interactions as needed, such as disabling cast and receive shadows. You can even render them with an alpha channel.
You may have watched this before, but I used a spherical render, mapped to a sphere in the sequences with the astronauts near the end. I wished I had hit on the method with some of my earlier robot character sequences. I rendered with an alpha channel and placed a star field image map in the scene's background.
https://www.youtube.com/watch?v=iS-sNje4k0o
When I was trying stuff a few months ago, I tried 16000 x 8000, and it was just taking too long for my liking. So I knocked it down to 12000 x 6000. That's still more pixelated than I'd like, so I'm not that interested in going higher. I'll probably go back to the smaller 8000 x 4000, and just keep the distant poles blocked from view.
It was during those experiments that drew me back into using Carrara's skies, but we can't use those for everything.
Anyways....
I kind of figured you were doing it for more than just speed. ;-)
If I may offer a possible suggestion, you could try and lower the resolution of the image and apply it to a scene encompassing sphere. Use the spherical image sequence in the sphere's color and glow channel, and limit the light interactions as needed, such as disabling cast and receive shadows. You can even render them with an alpha channel.
You may have watched this before, but I used a spherical render, mapped to a sphere in the sequences with the astronauts near the end. I wished I had hit on the method with some of my earlier robot character sequences. I rendered with an alpha channel and placed a star field image map in the scene's background.
https://www.youtube.com/watch?v=iS-sNje4k0oOne of my next experiments was going to go as follows:
* Set up my realistic sky and lighting solution, which includes a Sun Light for use with the RS atmosphere system - I love how they work together.
* Set up the North portion of the scene, fully animated for ten seonds
* Set up a mock bit of 3d assets to check lighting when placed NE, NW, E, W, S, SE, and SW ( duplicated assets at the same scene depth, parented to the camera, so that it remains in the render as I rotate the camera around), to make sure I like the lighting/atmosphere/etc., as I'll need to keep those things the same through the experiment.
* Set up a spherical camera at ninety degrees, to film horizontally, and set an appropriate height.
* Render the sequence at 10,000 x 5,000 (could be whatever you want) as a sequenced tga with alpha (alpha bit is another part of 'my' experiment)
* Load the resulting spherical animation into the Background.
* Set up an animation for the next chunk of real estate, making sure that its geometry overlaps the render from the last pass, without changing the position or rotation of the spherical camera, and render again.
* Repeat the last two steps until the spherical image is completed.
* Create a full scene that completely surrounds the filming camera to a distance away of 50 - 100 feet away from the camera, with a lot of gaps letting the spherical animation show through - looking for areas where we can take advantage of depth manipulation to alter what we see in the background even at very subtle angles. This scene will have a floor to it, so we'll never render the bottom third of the spherical map. If the alpha experiment fails, use something, be it a group of clouds, over-hang, or any other effect or geometry, to hide the ceiling of the spherical render.
Hide all portions of the final scene that will not be in the shot being filmed - using this as a repeat visit stage.