Home Creators Posts Import Register Favorites Logout

Content

A breakdown of 15 years of sticking Josh in sci-fi vehicles! Also a quick-n-dirty tutorial towards the end. This one talks about a fusion/resolve/blender workflow, but if you want to stay Blender-centric, this video is still pretty relevant.  Oh and here's the Fusion cleaned-edge-color tutorial I mentioned finding useful (also hah!! Forgot it used Tears of Steel footage :P)

Files

Compositing Josh in Cockpits

A breakdown of 15 years of sticking Josh in sci-fi vehicles! Also a quick-n-dirty tutorial towards the end. This one talks about a fusion/resolve/blender workflow, but if you want to stay Blender-centric, this video still covers a bunch of useful stuff. https://www.youtube.com/watch?v=RxD6H3ri8RI Oh and here's the Fusion cleaned-edge-color tutorial I mentioned finding useful (also hah!! Forgot it used Tears of Steel footage :P) https://youtu.be/xRkkUQpaYJg

Comments

Ephisus

Was half expecting to start on Spacemen belong in the future probably.

Roman Kovalev

How do you do reflections of characters and other CG elements in real footage? I know that there is shadow catcher, but no info about REFLECTIONS (or it is too complicated)

alex

Omg more pf this type of stuff! greensceen stuff😱DOPEEEE

Blake Rizzo

"After doing 800 VFX shots we found out the box could not handle it at all" *eyes twitch nervously

Paul Spooner

I'm still sore that the warning label texture links were broken in the 2:46 shots, so they ended up as ugly black rectangles. And then some of those textures worked at the 4:01 render, but others were broken! What is with the Arizona Suit cockpit that hates loading all the textures?

IanHubert

I realized mid-upload I forgot to give you a shout-out for the Arizona cockpit!! But dang! I forgot about the warning label issue! Although I THINK that was the project file where we learned that the bump mapping was causing all of those aliasing issues, too?? I'd had all these aliasing issues for the entire project, and never realized it was because I was pushing the bump maps too far, and I remain a little bit scared of bump mapping to this day :P

Griot1420

really excited to share this with someone, not great for all applications but has saved me a ton of time compositing. Image Sequence (color) into Diffuse BDSF into top slot of a mix shader. Take the alpha of the sequence and put that into a Principled BDSF, then into the bottom slot of the shader. Take another line out of it and invert it and put that to the factor of the mix shader. This allows you to relight the green screen element with the lighting that's in your scene. Since it a two dimension object it doesn't look great in every application, but it goes a long way in some. Good luck!

conorsvfx

great tutorial as always! I was wondering if it's possible to do colour correction in the shader editor of blender? I know you can use hue/sat and rgb curves but lets say you had a 3d model of a person you filmed so you knew exactly what colour your image plane should be in the scene. Is there any way of just eyedropping a colour from the model and applying that to your footage? If that makes zero sense or you've covered it already please lmk

Paul Spooner

And you never let me near a cockpit model again :P

Paul Spooner

You didn't talk so much about it in this particular video, but just wanted to give a general Attaboy on the improvement of your mechanical design sensibilities over the years. I've been doing real-life machine design for years, and you've managed to capture a ton of the ergonomic and manufacturability sensibilities that even engineers tend to miss. I can tell you've really been putting in the effort to not only look at reference, but understand the design constraints behind the geometry of everyday objects. Or, at least you're faking it convincingly, which is arguably even better. Inspiring as always!

conorsvfx

that link was VERY useful and i would like to marry you. Thanks so much Ian

Zeke Faust

It's quite interesting how rendering speed and capabilities of 3d have now advanced so much that it's changing a lot of the rote modalities that we associate with typical compositing or footage integration. There's a lot more parallelism between 2d and 3d now whether it's volume capture or the type of workflow you're using here. I mean obviously all the same color processing and image science still applies, but the process has become less segmented. I used to always err on the side of doing things in comp because it offered more interactivity, and to do another 3D render meant hours of waiting to make one change! But now interactivity is on the same level or even better than comp in some respects.

Kermit TheFraud4

wow this is a problem I have been having for ages, thanks so much Ian!

IanHubert

I KNOW! The degree to which faster GPUs/render optimizations have made the rendering process a lot less intensive has, like, fundamentally changed the way I think about CG. There's all this stuff now where I'm like, "why didn't I always do it like this?" and the answer's just "cause you really COULDN'T, until recently".

German Rudecindo

It must be really fun to be able to keep the render mode on all the time and move around like that. Still waiting for that water tutorial btw 🥲

Dustin Bowser

Happy to see you're playing around with Fusion. Give me a shout if you need any help. We use it pretty heavily on big stuff at our studio. https://jmp.sh/Ggj4Yn2 @Zeke Faust knows ;)

JTM

When Ian is in the Phaeton getting those slower render times he can understand how we feel when we go into cycles with the default cube.

chunky8bit

do you have a different version of camera shakify available? ever since 3.0 it hasn't worked for me, yet you're using it in 3.2 alpha!

alex

same here

alex

Hi i was wondering a few things. You dont use the lut tool you made? And you dont colorgrade the footage before adding it in to the scene? Why use an exr and not a png seq? Thanks!

Paul Wright

Great timing for this video. Thanks buddy.

Zeke Faust

PNG is much slower and doesn't handle alpha correctly or support floating point values! Also it can be way bigger in size than a compressed EXR. It was never really meant to be used in image sequences. Regarding the color, grading happens at the end. You want to keep your shots and elements as true to life as possible, then apply a creative grade after. This is because you might change your mind or want to make a small adjustment, and if the grade was baked into the render you'd waste time re-rendering things you shouldn't. It's best practice to linearize your footage when doing VFX to represent real life light values (check out Nathan's color videos for more info). In this case, the footage was converted from log to srgb/709 in resolve, then converted again to linear in the Blender shader editor

John Abraham

I CAN'T... that is incredible.

Bernhard VonZastrow

I remember working on a movie years and years ago (very late 90’s) where I had to put a greenscreened actor into a cockpit, and the ship flies backwards away from camera, us looking at the actor’s face the whole time. Because the actor was shot statically, the artificial pull back of the virtual camera made their face look weird as the shot progressed. The solution we came up with was to build a cg human (kind of a feat for me back then) and pose it to match the actor. I then projected the plate on to this mesh and it reintroduced the subtle parallax in the face that the camera move necessitated. I was kind of proud of that aspect of the shot. But I botched the rest of it so badly that I’ll never ever ever admit to having worked on it. Even if you pull all my fingernails off.

alex

Thanks man!<3

Nathan Vegdahl

As suggested by Zeke, you just need to download the latest version from Github. It's been working on Blender 3.x for a while now.

Nathan Vegdahl

The reason Ian isn't using LUT Maker/OCIO Maker in this case is because Resolve already knows how to linearize footage from Blackmagic Design cameras, which is what we shot on for this scene. So the footage is already being linearized (aside from the sRGB encoding) when exporting to EXR. (Ian glossed over how to actually set up Resolve to linearize things, though. Which is understandable, because getting Resolve to do it properly turns out to be... a little tricky. Resolve tries *really* hard to apply filmic s-curves to everything, and turning that default behavior off is buried under mulitiple levels of menus. But it is possible.)

chunky8bit

yes, it works! thanks a lot you two!

Arsh Waseem

can't wait for more of this this has always been an issue for me

Carter Burr-Kirven

Now I'm curious, I always just use the Color Space Transform and I thought I was doing it right! I do have S-curve contrast turned off in my settings too, but even then I thought that only mattered if I was using the Contrast slider, which I'm usually not, just passing plates straight through the conversion and writing to EXRs. Is there more I need to be doing? I'd hate to realize I've been inaccurately linearizing my plates all this time, haha!

IanHubert

AH YEAH! That makes sense. (ALSO- 1000-to-1, but if it was Lost In Space, you HAVE to tell me because that is my legitimate favorite friggin Bad Film) I've actually spent the last 2 days doing that (for a video call in the scene, I'm trying to have a holographic head, and I figured there was some "easy" way to track the 3d head to the footage, but I ended up just brute force tracking the whole conversation so I could project the video texture onto a 3d model of his head... it works okay, but only because it makes sense that the "in world" tech is super janky :P) OKAY NO BUT SERIOUSLY was it Lost in Space?

Bernhard VonZastrow

Name, rank, and serial number is all you’ll get from me! I am willing to accidentally let slip that I modeled the “actor” using form-Z (look it up whippersnapper) and did the projections using electric image. Well, since it is just the two of us I’ll also divulge this much. It’s not Lost in Space. I do remember watching that film in the theater and for some reason I seem to recall that they did bullet time in it? If I’m remembering correctly (something I wouldn’t bet my own money on, but maybe if you lent me some?) then that would mean they did it before the matrix. The reason I seem to recall that is because a coworker of mine asked me how I thought they had done it and whether we could do something similar. At the time the only way I could imagine was a variation on what we were discussing: a 3D match move of the actors and projected plates - though way back then that would have been a TON of work.

Zeke Faust

Carter - The CST method is fine as long as you disable the Forward OOTF checkbox. The backwards/forwards OOTF settings are for going from display to scene referred or vice versa. So you want both boxes unchecked. The other method is to use Resolve Color Management with these settings: Color processing mode: CUSTOM Separate Gamut/Gamma: Checked Input color space [whatever camera input] Output color space: Linear And *most importantly*, change the Input and Output DRT to NONE. This setting is basically a tone mapper which gives the highlights a shoulder which obviously you don't want. And yes, you're correct that the S-curve contrast setting only applies when adjusting the contrast control! That's separate and has no effect on the RCM settings EDIT: PS: FYI, IIRC the CST also has a tone mapper which you want to turn off

Carter Burr-Kirven

Okay, phew, yeah, already doing all of that! I was just worried, thinking Nathan made it sound like there were some other hidden settings messing with the math, like from doing color space conversions inside a non-color-managed project or something? Seemed plausible, y'know. I don’t think I’ve ever tried doing my plate conversions with RCM instead of manually… maybe I should give that a go sometime

Carter Burr-Kirven

They did indeed! The "time-slice" effect, as it was often called at the time, was big in commercials throughout the 90s and showed up in a few movies, like Lost in Space (1998) and Wing Commander (1999, just a few weeks before The Matrix!) before making it really big as "bullet time." The sequence in Lost in Space looks real cheesy, I don't know if they shot it without enough cameras, or faked some or all of it with rigs/wires, or what (they also got the shutter speed wrong, so the actors are frozen but also still motion-blurred), but Wing Commander looks pretty good https://youtu.be/UnDOTYYkTyU

Carter Burr-Kirven

Update: Wing Commander used an array of 120 cameras, while Lost in Space used only 10 and some early morphing software (source: Cinefex). So... that explains that!

Kyle Sullivan

I've run into an issue when dealing with imported cards that I'd love to get your take on - I'd like to be able to adjust the exposure in Blender to match the exposure of the video element, but with the video placed in the scene, it also gets brighter with any digital camera changes. Is there a way around this, or is the answer just to not touch Blender's exposure setting?

Carter Burr-Kirven

You have a couple options there. Blender's exposure dial is, by its very nature, a global adjustment, so it'll always affect everything in your scene. Your approach might be a little different depending on whether your image plane is self-illuminating or a PBR material existing within the scene, but either way you're generally going to want to A) add more light to the scene to actually brighten it, instead of just cranking up the universal knob afterwards, or B) *do* use the exposure knob for a global adjustment, and just compensate by reducing exposure on your image plane material. You could also split your render into passes for the person and environment separately, letting you control color on each individually, while still keeping them indirectly in one another's scenes for shadows and reflections and stuff.

IanHubert

Carter’s totally right!! With this workflow you probably want to adjust either the lighting in the scene to match the footage, or the footage (in the shader editor) to match the scene, but unfortunately any exposure adjustments would affect all-of-the-above

Bernhard VonZastrow

Aha! Memory vindicated! Thanks. I had wondered about that for a while but it would have required that I “do work” searching for “information” to validate. Much easier to just keep wondering and wait for the day when some random internet comment would cause me to mention it and then some kind and intelligent stranger would do the research for me. A strategy that works every time (if every time is defined as once every ten to fifteen years).

Nathan Vegdahl

Ah, sorry. Yeah, I didn't mean to scare you there. I haven't actually worked with Resolve much, so I'm coming from the perspective of "guy who knows more than a thing or two about color, but is largely ignorant of Resolve". For someone like me, Resolve makes it frustratingly non-obvious how to do technical color transformations without Resolve stepping in to also try to make things "look nice". But if you already know where all the boxes to tick/untick are, you're probably fine. (As an aside: Ian and I actually mistakenly thought we *were* linearizing the footage at first, because we found what seemed like the right settings in Resolve, and things looked linear. It wasn't until trying to reverse-engineer BMD's transfer functions (https://psychopath.io/post/2022_04_23_blackmagic_design_color_spaces) and viewing the resulting graphs that we discovered Resolve was still doing other things we hadn't asked it to. And it was only through that that we finally figured out how get Resolve to properly linearize things. Mind you, those additional things probably make sense for colorists, so it's nothing specifically against Resolve. It's just frustrating when you're trying to do technical rather than artistic color transformations.)

German Bauer

I never thought we'd talk about interlacing again...totally forgot about this. PAL...SECAM...NTSC what a nightmare...thanks for the retro vibes

Zap Andersson

So I'm watching "Severance" on Apple TV. The lead guy looks EXACTLY like Josh Truax

Siavash Robb

can you do a video on how you get the nice water

IanHubert

I KEPT THINKING THAT THE WHOLE TIME! Also- such a good show.

Pete Burges

Man, that time-difference thing and redoing your fx shots sounds nightmarish...and somewhat familiar.

Bo Maryniuk

Been doing exactly the same workflow! And also the same tools (DaVinci etc). :-) Next step for you: "How the heck do I do light wrap on that pre-keyed footage!?".

bpype

Hi Ian, is the project london assets still available? I remember download it from the website

IanHubert

I think the Goose might still be up on Blendswap! No idea where the other stuff is at this point (it’s all like 12 years old, too, and most of it doesn’t work with cycles or anything :P)

IanHubert

Yesss- though honestly, prohibitive render times aside, a bit of volumetric haze gives you a lot of nice interaction! I can’t wait for one of these newfangled AI solutions to let me subtly re-light the footage, too.

Bo Maryniuk

I still have a hack for it. You can pre-lightwrap it in davinci, having pre-rendered CG part from blender on background, and then composing it there for mere light wrap for the footage. Then put the footage back to Blender as you do, and it works quite well, apparently. Just don't overdo it with the virtual camera, so then the light wrap doesn't match that well anymore. 🙂

Tyler Adams

Great video! I hope you can visit HDRI content like De-Lighting a photoscan and projecting an HDRI onto an environment scan for more accurate reflections/compositing! Thanks for the constant inspiration!!!

Zap Andersson

I'm not even kidding. I had just come from watching your video, and went into first episode of Severance and I was like.... is that.... no.... he looks a little too old..... but... IS IT? Is this Josh from 5 years in the future? Or is it coz the footage you showed up here is 5 years old, and it really IS Josh? But I slowly realized it probably isn't, but on first sight I was almost falling off my chair....

Hans Jacob Wagner

Is it possible to extract a depth map from a single image or video clip in blender. Like what the new fusion hype is all about?

IanHubert

Oh! Well- you can definitely render out a specific depth pass, but I don't think Blender will generate a depth pass for random video files, unfortunately.

Hans Jacob Wagner

I get that you can render them out :D But it was more the possibility to add image planes into a stop motion scene using depth map masking rather than using greenscreen. I will see what Midas can do. Just wondered if this kind of ai masking already had been introduced in blender.

Juniper Ysadora

Really great video also something that is actually completely and utterly unrelated is; have you considered doing some sort of podcast or something like it either on Patreon or somewhere else? I could list all my reasons for why I think that would be a great thing but instead I will simply leave it at the question!

Noneya D Biznazz

seconded! would watch :) I think the long stream of consciousness style creative vids are my favorite, you never know when he's going to drop a nugget of gold and speed up your workflow 20% or show you something you never even knew was there

Marcus Sidonio

Ian did you rendered the final scene at cycles or evee? what is your setup for rendering scenes and what passes you usually export?

Sam Dunstan

Sick. Also, loved the interlacing rant!

DoDzi

Hi, have you considered UE5 character lighting workflow? Creating a scene (let's just say a spaceship interior) almost fully in blender with built in lighting, lamps, etc. Then exporting the meshes and lighting into Unreal Engine. There, using the live production method of keying out character green screen footage live while filming, and seeing it in Unreal with the 3D scene. Wouldn't that lead to more accurate character lighting in the scene, since we would have live reference of the lighting? And there would be less need of creating different point lights for every shot to match the green screen. I am trying to create an almost full CGI short film of my own and was thinking of using this method. Would you have any thoughts of it? Also CHEERS! Amazing tutorial as always!

Mickey

LOOKING FOR: 1 - a IAN hubert tutorial focused on, placing green screen elements as ( movie planes ) in 3d space and adding shadow catchers. 2 - anything focused on color correction. Working on a big project. I APPRECIATE ANYONES RESPONSE

Jeffery F

Hey thanks for the great tutorial, Ian! I was wondering if I export my log footage as an exr sequence, does that automatically make it properly linearized?

Ben

Is there a more structured and detailed version of the “wild tricks” video on here? I love that video but Ian really breezes through the technical parts I need to understand to do it correctly. In other words, the way I understand it, there are 2 ways of blending footage into a 3D world: 1 - your real camera is actually moving around your subject, you 3D track it and that becomes your blender camera. Boom, job done, layer how you want. 2- you shoot your green screen element static, and move your blender camera to your heart’s content *as long as* you stay pretty front on to not show that it’s a 2D element. But his wild tricks method is like a blend of both? Parenting the greenscreen footage to the camera, and also doing the foot-at-bottom scale trick? I guess I'm just not really understanding. Maybe there's a video already on here that goes into this in detail? I'm pretty new on the patreon so maybe I just haven't seen it. Thx!

Sascha Rossier

i can't take my eyes off those intricate details in your cockpit ... its a absolute masterpiece ... the Makers of "Space Cowboys" could learn a lot about ODO (obsessive-detail-obsession) from your tutos ... Masterpiece!