- Aug 31, 2020
- 551
- 7,002
- Feb 1, 2018
- 65
- 279
- Mar 23, 2020
- 65
- 445
I'm not so sure that he is using a crowd package, I don't recall any of them having Halloween party theme (I could be wrong of course).
Other options available are that you can create your own billboard characters and use them in a scene, I've taken that approach before in the past.
I also noticed in the images shown that there were no shadows being cast, so I suspect that in this instance it is just a case of layering images. Render the background, and then render the characters in multiple passes without a background, and then just layer the images in whatever post editing tool you decide to use. I've used this approach with animations where I wanted to have a lot of characters moving about within an environment; Although I layer those rendered files using Davinci Resolve, but the same approach can be performed in Photoshop or any app that supports layers. If the characters need to cast shadows, then that can be achieved by using an Iray Matte. Or alternatively just add shadows in Photoshop as part of the post process work.
Couldn't you catch the shadows on a canvas while rendering the billboards? One by one, then use the same technique to layer them in photoshop or your post editor of choice?
- Mar 23, 2020
- 65
- 445
- Dec 25, 2018
- 298
- 7,058
- Nov 20, 2019
- 175
- 371
hey, where are the vaginas ? is this a public version
?just kidding
. btw amazing piece of art.
- Apr 14, 2019
- 3,605
- 39,807
- Apr 21, 2018
- 280
- 4,663
- Jul 20, 2018
- 701
- 1,158
Hey, Thank you!
I'm far from an expert and maybe others have better ways of doing things.
These renders took quite long as I used canvasses to render each group of people (but I do have a quick and dirty way below). So about two hours for me with a 1070 and 2070 running together. I did see that I could have shortened it a little as you don't see that much difference between 80% and 100% (at 95% convergence). And I have some issues with my PC at the moment that cost 10 - 20 minutes per image of render time easily where my pc sits idle and looks like it crashed (unable to move mouse) before the render starts.
I first rendered the background and already set the cameras. (make sure to lock all cameras you use in place or you will have a bad time.)
I made each group of models their own file. So to keep track of the overall composition I used a cylinder primitive to see where a group was. (though next time I'll use two of them One for the main body that blocks the camera and one larger that also encompasses all extremities
Now (two previous groups in this case):
next time:
To use canvasses to render the image I followed this thread.
As stated I do have a quick and dirty way that speeds up rendering to about a minute per render (excluding loading times for the scene)
you can use an HDRI of the background to light your scene to get a similar but slightly bland version of the scene. (Though you might be able to get a better result with tweaking the HDRI in photoshop).
I did a post about it long ago here. The issue with the ground I mentioned can be fixed by setting the ground to manual btw.(This has advantages when making animations as you can render a long animation in a couple of days rather than weeks)
With HDRI:
With canvasses:
I hope this answers your question a little and helps you make your own scene.
Yeah I only noticed the shadows missing when I was composing the images together. I used this technique for the first time have a few things to learn
Yes I used Krita in this case to compose the image.
Last edited:
That was not a critique. I was simply answering BravoMDB's comment. I just thought catching the shadows with a canvas or five would be easier than the methods he mentioned. As for your render...I know I wouldn't have the patience. Not with my 970m anyways.
- Feb 21, 2018
- 2,062
- 21,772
- Jul 20, 2018
- 701
- 1,158
- May 2, 2017
- 183
- 1,364
- Jun 4, 2019
- 3,388
- 13,598
- Jun 12, 2020
- 1,461
- 22,342
