DaVinci Resolve: Using depthmap to convert 2D video to stereoscopic 3D

DaVinci Resolve has a new feature as of version 18, depthmap!
Now you can convert 2D videos to stereoscopic 3D almost instantly.
patreon.com/WheatstoneHolmes

[@kellyluck1626]

Hot dog! Just used this to do a test vid, then modified it to do half-width SBS. It definitely takes some time to render out, but I'm very pleased with the results. Thanks for this!

[@WheatstoneHolmes]

Awesome! Glad it worked for you!

[@p.m.3437]

Hey, Thanks for the great video! It's wonderful to see that a new feature adresses stereoscopic 3D, since 3D was said to be dead already. Well, you'll never get the quality of native 3D, but it's a great start to include some 2D clips in a stereoscopic project and give theses at least a little bit 3D.

[@Meteotrance]

Also separate the compositing element in real 3D Space is the key to make converted 2d depth map convincing, on most movie converted they lazy on that process, on titanic 3D convert they separate and rotoscope every elements, and place along z axis to realy add sense of depth.

[@bigipalabom]

a really awsome video, thanks for sharing the techniques, I just bought a pair of vr glasses, can't wait to convert my drone footage to stereoscopic!

[@WheatstoneHolmes]

Thanks!

[@ShadowSnake141]

This is interesting! thanks!

[@sayinglikeitis3263]

Could you do a more in depth tutorial on this. I spent hours trying to find info on stereoscopic 3D but there really isn't anything straightforward.

[@WheatstoneHolmes]

I'm still learning more about S3D in DaVinci Resolve myself but I'll see if I can make a video that is easier to follow. Thanks for the comment.

[@sayinglikeitis3263]

@@WheatstoneHolmes That would be awesome. Thanks in advance.

[@TheUpscaleTech]

​​@@WheatstoneHolmesyes please if you can, thanks because there is nothing out there on how to convert 2d to 3d

[@WheatstoneHolmes]

Have you tried 3DCombine?@@TheUpscaleTech

[@TheUpscaleTech]

yes i have tried it but i prefer owl 3d over 3d combine. Was looking for an indepth video of how to do it in davinci because no one has a tutorial in davinci and i wanted to see if doing it in davinci would be better. Were you able to achieve better results with davinci over 3d combine?@@WheatstoneHolmes

[@Eincrou]

I've known of DaVinci Resolve, but this is the first time I've seen it in action. The node editing feature is very interesting, and simplifies what I would have to use nested sequences to accomplish in Premiere Pro. It's amazing to see a video editing software add something new related to stereoscopic 3D.

[@WheatstoneHolmes]

The depthmap feature is supposed to be used for color grading by depth but I saw someone make a depthmap by hand and when the new feature came out I knew I had to apply it to S3D!

[@Eincrou]

@@WheatstoneHolmes Ah, so this is not an intentional S3D feature, and is a "clever use of game mechanics" situation. Well, that mentality is how I have been able to make hundreds of hours of stereoscopic 3D video content in Premiere Pro despite it having zero support for non-VR S3D editing: Clever use of the ability to nest sequences. You'd think the various NLEs claiming to be the most full-featured, professional editing software would all want native S3D support. Or at least simply to not fall behind a 3rd rate NLE like Vegas. But no, we have to be clever or buy a $400 plugin.

[@WheatstoneHolmes]

@@Eincrou LOL! Thanks! Blackmagic Design put a lot of stereoscopic features (including VR) in the software but for some reason depthmap is advertised as a color grade tool. Fusion Studio is another of their software offerings and it supposedly allows for editing while wearing a VR headset!

[@ji3200]

Is this better than 3d combine software? I wish subtitle work since i love japanese animation

[@Eincrou]

@@ji3200 I have no idea what you're talking about, sorry.

[@robertshaw88]

This is great. Wish there was a in-depth tutorial or template for doing this. as i'm new to DaVinci Resolve. The only other way that i know of converting 2D video to stereoscopic 3D video is using 3dcombine. Anyone know of any other user friendly software for doing this?

[@WheatstoneHolmes]

I think 3DCombine is easier than this method.

[@totox691]

Shorter, use only one renderer3D and check stack, so you dont need the combine and second renderer node

[@Music.Contest]

Hi. Everything looks good when I preview media out on fusion page. When i go to edit page or deliver page, the video only shows the depth map not what i see on fusion page. How do i render out to a stereoscopic mp4 file for side by side? Thanks

[@rwernyei]

Could be a couple settings. I reached out to your wakingself account.

[@NUCLEARARMAMENT]

Is this method more practical and user-friendly than using MiDaS AI depth map generation with Stereo Photo Maker to combine stills into a final 3D-upconverted video sequence? I'm wondering if this method would produce the same results within Resolve, as it's much more streamlined, but I still wonder if it would produce inferior results to the aforementioned alternative.

[@kellyluck1626]

I've tried the SPM method and it does work, but I think this does a better depth map. And of course you don't have to do all the muxing & demuxing and so on.

[@inv1cmaj]

This is very cool. How would this be delivered to a format that can be played on a 3D TV or projector with the active 3D glasses ?

[@WheatstoneHolmes]

At the end of your flow you would render as top-bottom or side-by-side instead of anaglyph, most TVs and projectors and even VR headsets can use those formats.

[@MarnikWillems]

Is it also possible in DVR 18 to export for an active shutter 3D system?

[@WheatstoneHolmes]

An active shutter 3D system such as a TV or projector usually supports side-by-side or top-bottom 3D formats so if you render your final video in one of those formats from DVR it should work fine. In this video I go show the formats: kzbin.info/www/bejne/hoSQaqd-l5uUq9k

[@martinhansschmittcom]

Could you post a screenshot of the Fusion workflow, please? Or make a tutorial about this?

[@clamo6362]

try this instead. media in -to- imgplan3d -to- 3dmerg.....cam on 3dmerg, 3dmerg -to- displace.....depthmap on displace, displace -to- render 3d. displacement settings....point to cam. then adjust the cam as well as the other adjustments you need.

[@WheatstoneHolmes]

Thanks! I'll try that!

[@warrencapes8286]

I can't get my camera to align to the Merge3D - my Merge3D is outside the yeloow box - how can I quickly align it ?

[@WheatstoneHolmes]

I'll have to look at this effect in DR again to refresh my memory and then I'll see if I can get a solution for you.

[@rwernyei]

In Camera3D1 node Select Transform>Translation change Z value. That should do it.

[@pastuh]

Maybe exist full "normal" tutorial? :X

[@salaciouscreations4323]

The fusion nodes never link up for me.

[@OurResistance]

All I wanted to do was take a 2D image + depth map to get an SBS image! Heck, I guess I will have to try to make my own software for doing that!

[@WheatstoneHolmes]

Try my other video, it might give you better results. kzbin.info/www/bejne/hoSQaqd-l5uUq9k

[@rwernyei]

I personally like the DaVinci/Fusion method better for more manual control but if you want fast easy results, try Owl3D. To my eyes, Owl3D's AI creates better depth maps than Resolve.

[@cinematricks]

cool

[@MrCarstennielsen]

is this real 3D, I mean, 2D converted into 3D? I thought you needed 2 angles to create this, meaning 2 linses recording a video, but I reason, that it would be possible to create very good results, espc. when video is created only from steadry and set position at a rather long distance from the objects in focus, espc. if there is no need for real closups, I mean, a boxingfight would be great to create a 3D version of; Is it really possible to do so it is very similar to real 3D? I am intersted only in polarized 3D I guess, as this is passive I have acesses to by my passive 3D tv s. I have active too, but that really is no use.

[@JT-mz6ep]

Depth-based image rendering is KEY. Check this 2005 paper titled "STEREOSCOPIC IMAGE GENERATION BASED ON DEPTH IMAGES FOR 3D-TV": www.researchgate.net/profile/Liang-Zhang-33/publication/3041153_Stereoscopic_Image_Generation_Based_on_Depth_Images_for_3D_TV/links/09e41511bb0d2d62e8000000/Stereoscopic-Image-Generation-Based-on-Depth-Images-for-3D-TV.pdf

[@WheatstoneHolmes]

This is just a method of converting 2D video to 3D. The results can be rendered as anaglyph (red/cyan), side-by-side or over/under 3D video. Most of the video I created I used 2 cameras as you said or 3D cameras with 2 lenses. My polarized TV can display side-by-side, over/under and Checkerboard, but I don't see anyone making checkerboard 3D.

[@gentlemood]

I thought this was going to be a tutorial but you're just showing up. Only advance people that could figure it out on their own can understand it.

[@rwernyei]

@WheatstoneHolmes I was going about it all wrong in DaVinci Resolve Studio 18. I used the Depth Map feature with the Color Page to create Anaglyph 3D. Here's my result: kzbin.info/www/bejne/kHK1cpZ_oLl_Z7s By following your video and making some tweaks in Fusion, I made this: kzbin.info/www/bejne/rGmvioCjrN13q9E Much better, Thanks!

[@WheatstoneHolmes]

Thanks for sharing your results. I've been very busy for months now or I would do some more work with DaVinci and depth maps. I watched the clips and in some ways I think the Color Page version was better, maybe it was just my glasses.

[@rwernyei]

@@WheatstoneHolmes I think the Color Page way keeps more detail and creates a pronounced effect on subjects in the foreground but has no displacement making them flat. More like stacked 2D imagery. I like your method because when you tilt your head side to side the image moves much like the old Cracker Jack 3D cards.