|
AI reconstruction in general is great & becoming really important in what we do. The last commercial project I posted was upscaled from 24fps/1080p to 60fps/4k and it looks spectacular. Definitely not as good as a native 4k60, but considering the source, it looks significantly shaper and smoother on a TV after an overnight semi automated process.
|
#
¿
Feb 14, 2021 08:32
|
|
|
|
| # ¿ Apr 25, 2024 11:23 |
|
|
Comfy Fleece Sweater posted:Getting into this 3d thing right now seems like a really weird, possibly bad time to do so. Because of the AI/Procedurals/MachineLearning explosion. Automated tools will never be perfect - they'll always need some level of human touchup. There will always be a need to understand what good topology looks like - it's only way to know why something is breaking, and what needs to be done to it to fix it. Much like everyone worried that content aware fill would put the entire retouching industry out of business, it didn't really because good retouching is more about the decision-making that goes into it, rather than the tedious grunt process of actually doing it. And when the automated tools fail, you'll still need to touch it up by hand. That's what humans are required for - good decision-making. Learning the tools and processes' tends to give you a very natural and deep ability to make better decisions - it's just hard to quantify over a short period of time. We've actually had a bit of an issue in arch/marketing with people who've only ever worked using premade downloaded models and premade materials - the second something requires complex, custom work they are completely incapable. You absolutely need to know all of those things still - if only to go through the process to learn pros, cons, and develop a deep familiarity. And when to spot things and clean them up before they become issues that are harder for others to find. AI & machine learning is saving render time, allowing the quality bar to be raised even further. Megascans assets & these human tools allow people to jump straight to layout and skip all the boring poo poo that would've taken months to model. I'd actually argue we're in the golden age now. The time to go from making a decision and realizing it is shorter than ever, allowing artists to push the quality and standard higher and higher. Single people can render personal projects in crisp 4k on a home machine and have access to all the tools they need to make it look close to what large studios can put out. I've been doing this for 18 years now and (tools and processes wise) every single year has been better than the one before to be in this. It's always moved this quickly. Techniques and processes get replaced constantly, just keep yourself in front of them and adapt. I have seen many people left behind by thinking they'd learnt all they needed to and could coast as a pair of hands who's entire offering was knowing what buttons to press - by the time they realized they had left it too long, they were almost starting from scratch.
|
|
#
¿
Feb 18, 2021 18:43
|
|
|
justcola posted:Is there a big torrent anywhere of textures? I want to start over as I'm so badly organised and have a folder with stuff thats different sizes or doesnt tile correctly etc. I'm aware of the free texture and hdri websites, but was wondering if there was a way to grab lots quickly than downloading them one by one. I think a freelancer subscription to poliigon is the cleanest way to go. Many websites that sold textures that have been packaged up on torrent sites are either ultra specific (arroway etc) or so general and of such a mix in quality you probably want to avoid using them anyway. If it's for hobby work & you're dead set on not paying for them, get megascans and do the pinkie promise that you'll only ever use them in UE4. NFTs are stupid and i cant wait for it to all crumble. Had a long discussion about it on facebook with some other artists this morning. It's a giant pyramid scheme - the amount of lazy, quick digital art being generated just for sale is saturating it already - supply is far outstripping demand, while the guys running the service are taking cuts from everything listed for sale. art being valuable is also very closely linked to it's rarity, and if the 'artist' is uploading 40 new pieces a week, original purchasers won't be able to find a market to resell to making their original purchases worthless. It's only worth anything if you can find someone below you and this is why scarcity is essential. If scarcity is irrelevant then just sell prints. cubicle gangster fucked around with this message at 20:30 on Mar 3, 2021 |
|
#
¿
Mar 3, 2021 20:24
|
|
|
Did you expect anything else? I was convinced it was Elon musk that bought it. He bought Bitcoin this past year ending up with a 1.5bn stake. Grimes had sold something as an NFT, I was convinced it was him loss leading to give it legitimacy. I strongly dislike all of this and am praying it all spectacularly burns down soon. Any artist that sells a piece for 70m immediately values everything else they do and have ever done in the 40-50m+ range. Is nobody going to sell a beeple again? Do we agree some beeples are worth more than others? The entire economy of this is beyond hosed. It's a little sad a legit source like christie's were a bit of a vulture to get a payday.
|
|
#
¿
Mar 14, 2021 08:41
|
|
|
Anyone here been responsible for leading/managing someone they hired this past year - going from a previous in-studio environment to introducing someone new to a full work from home setup? And on the other side - has anyone joined a new studio in a fully work from home setup after having previously been used to working in studios? we're going to have to do it soon and I'm curious if there's any insight people could share. From the person hiring/managing it would be good to know the kinds of hiccups you faced, and from anyone who started work in that kind of environment, what you felt could have been done better during the onboarding and how your workload was managed. My natural instinct is that it's going to be much more difficult to introduce people who have a freelance/solo heavy background - we have hired people in the past who exclusively work in-camera, dont use xrefs or layers, and do very little collaboration. organization for collaboration is always the biggest learning curve for someone who hasn't done it before and requires the most hands on in-person time, so that part kind of terrifies me.
|
|
#
¿
Mar 24, 2021 17:52
|
|
|
Jewel posted:studies show most salaried employees get 3 hours of work done in the average 8 hour day; pre-covid. it's fine. lol I want to see this study. was it souless government beurocratic jobs? I cant imagine anyone working in CG only putting in 3 solid hours of work each day. I think if you compare my best days to my worst, I've definitely got a full days worth of work done in about 3 hours once or twice because everything just clicked and fell together, but I wouldn't flip the comparison and say those other days I'm only getting 3 hours worth of work done. Really appreciate all the responses to my question. we've got a pretty rigid routine in place - a morning check in on video call at 10am to brief work and go over whats in the pipeline, then another optional call at 4pm which we'll have if we're closely collaborating and skip if everyones got a meaty amount of work to do. We use zoom for everything right now, and slack/google sheets for work reviews that need a little more documentation. We do need to create a workflow bible - how the server is organised, how we organise photoshop files, settings/elements we use, layers and general scene organisation/xrefs etc. That's probably going to take a while to put together and be a lot of work, but I imagine hiring someone without it would be a waste of everyones time. When we've hired in person, typically everyone that has been with us for a while will share the onboarding workload - one person will cover photoshop and sit with them for a few hours etc. I guess it feels odd to make it so formal - because now we need to block out a chunk of time and say 'youre doing photoshop while this new person watches your screen for 2 hours', even though it would always turn into 2hr sessions before. I have found that my days are so fragmented with phonecalls and emails, that sometimes it's better to just get stuff done around the house while keeping on top of things with work and set aside a solid uninterrupted 3-4 hours at the end of the work day when i get to open max and work with music. it's rare i get to divide it so cleanly though - usually what happens is I end up sat at my desk being interrupted constantly for 8 hours then having to do 3-4h of production at the end of the day anyway. cubicle gangster fucked around with this message at 21:34 on Mar 25, 2021 |
|
#
¿
Mar 25, 2021 21:03
|
|
|
Biohazard posted:Good advice! It's funny how important it is to have other people look at stuff after you've been staring at it for days at a time, you just miss really obvious stuff like that. Here's the updated "finals" i delivered. There's some pretty glaring things I wouldn't mind fixing, but I'm gonna call it quits on this one. Feedback is welcome though! I was meaning to give you feedback on this sooner, but I had my second pfizer shot late last week and was out of action for a few days! sorry that this is a bit late, but it should be pretty universal. I think one of the most important things to think about is contrast and levels and how that helps the viewer understand the lighting. Good architectural photography works just as well in black and white as it does in color - so a great practice is to turn your renders black and white to review them. Our process is to take early draft renders, do a rough as hell paintover - switching black and white on and off, while pushing the hue and various levels around until it feels comfortable, then we go back through the layers studying what we did, what it did to the image, and what we need to do in 3d in order to minimise or eliminate the need to do those changes in photoshop for the next round. I hope you don't mind, but I did a rough edit to your image to illustrate this. The purpose of this is to show you my self-review process, as it would be near impossible with text alone. This is your original image with the saturation turned right down. Key things to look out for are - you have some bright points in the shadows that are as bright as bright points in the sunlight, and similarly, dark areas in the sunlit area as dark as the ones in the shadow area. When all color is removed, it's a little hard to read the time of day - it kind of looks overcast, but then the hard shadow lines on the front face fight that reading a little. The grass is as dark as the underside of the dark metal awning, the interior ceiling is brighter than the sky. Having no colors on makes it much easier to spot all of this.  This is a super quick paintover I did. It's very rough - messy soft brushes, really aggressive curves, whatever weird techniques. the idea isn't to make something we'd show a client, it's to figure out what we want to do to the image for real at some point in the future. Key moves were lifting the grass up, making it feel much more varied with some darkening in the corner and on the right in the shadow area of the building. Toned down all the whites on the facade in shadow, and lifted all the black levels from the facade in sunlight. Pulled some of the detail back into the sky too, so that the overall levels of the image didn't have any major spikes. Really just focused on making the volumes and lighting read more clearly.  With the colors tamed a little it looks like this - I shifted the grass hue to be more yellow, took the yellow out of the direct sun and made the white balance a little more neutral.  The intention of this was to illustrate a way of thinking about the overall look of the image - I'm not saying this is what your image should have looked like. I only spent a few minutes in photoshop pushing things around to touch on the general thought process - once we're through this and we're finding the voice of the image and feel comfortable with the direction it's going in, we might look at making some changes to the lighting or materials. it's generally just a good process to use for self-review. Sorry if that was muddled, I'm still a bit foggy. cubicle gangster fucked around with this message at 19:22 on Apr 26, 2021 |
|
#
¿
Apr 26, 2021 19:18
|
|
|
Not doing as much of it as I'd like, but I'm still trying to come up with potentially useful, practical tyflow setups. https://www.instagram.com/p/CPQW232sRUV/?utm_medium=copy_link It was designed to be easily reusable but the density of the particles in the stems/leaves changes the mass so it always needs tweaking a bit. Only really works on big leafed plants too.
|
|
#
¿
May 24, 2021 14:21
|
|
|
EoinCannon posted:Made a catbus This is wonderful. I've not been doing much I can show recently, but I did finally start working on a tutorial that i've been talking about doing for years. It's aimed at people who currently mostly do post production in after effects - maybe they are familiar with how nodes work, but dont really realise what it is that makes them so powerful. It's a tight hour, starts off covering footage cleanup and painting things out like cranes without using any tracking, rotoscoping or hand animation, then it lightly covers using those same techniques on finished CG shots to do some matte painting type additions/fixes. I've got one more recording session left to go, but it's almost done! I have found the whole process extremely difficult, but it's been satisfying. Recording myself and attempting to speak clearly, without fidgeting, hitting a good pace, not saying 'UM' etc. It's a total pain in the rear end.
|
|
#
¿
Jun 23, 2021 23:46
|
|
|
Listerine posted:Are you selling it or putting it up for free? It'll be free. It's mainly because most studios in arch vis that do an aerial or a camera tracked shot (including people at dbox), you can see hand animated masks sliding about all over the place and paintouts shimmering, it drives me loving mad. nobody should be keyframing masks frame by frame for things that dont move! Figure it'll probably be valuable for a lot of other people too. Quite likley people who work in comp day in day out using nuke are going to think the way i went about it was kind of stupid, but hopefully some will find it useful. e: actually, thats a great point - anyone super familiar with nuke want to skim through it? It's in fusion but from my understanding this is all basic tools. i could still add an addendum of alternate approaches. cubicle gangster fucked around with this message at 01:20 on Jun 24, 2021 |
|
#
¿
Jun 24, 2021 01:08
|
|
|
If the interior volume is continuous, can't you just delete it? The interior hollow volume should be a surface with the normals flipped and inside another volume. Ive worked with rhino models that had cavities handled like that.
|
|
#
¿
Jun 26, 2021 18:49
|
|
|
Where's the best place to put an ad out to hire someone/a team proficient in UE - blueprints, frontend, interactivity? From a games experience point of view, is this considered UI? I took a look on Gamasutra, they have a contractor list 1700 strong but it's a bit of a random crapshoot to click through and try and figure out if they offer the services we want. We'd like to find someone (or a team) that we can potentially rely on for more of this - someone excited about the possibilities and experienced enough to suggest new approaches and ideas that we might not have considered. We'd supply scenes organized however required, and we'd be able to handle design elements of the UI if needed too - it's largely the programming and interactive component we need help with. e: here's a quick reel of some of the past experience spaces that we've designed & built - we get healthy budgets to do it properly. https://vimeo.com/341597152 cubicle gangster fucked around with this message at 17:10 on Jul 20, 2021 |
|
#
¿
Jul 20, 2021 17:07
|
|
|
I am pushing hard for a vendor we can partner with - I think long term it will make more a more collaborative process and both sides can benefit. It sounds like we should have a chat....!
|
|
#
¿
Jul 20, 2021 21:39
|
|
|
I've got a quick UE question - I've got my main persistent level and a sub-level with a shitload of geometry in. Can I move/scale the sub level, but only in one level that references it? I'd like to take a level that someone else can work in and detail up, link it to a dummy object, then in a different scene scale it down to 1/10 and move it. Imagine taking a scene you can walk around in at actual size, but also using that same level as a miniature model on a table. I know you wouldn't do it like this in a game because you'd have unique LOD ready models, but for this purpose optimization is irrelevant - both the actual size and the miniature version need to reference the same level. I've tried grouping the contents and linking an empty actor to it, but it seems I cant link the empty actor to the contents of a different level. I've switched to world composition, which gets me some 'level bounds' objects, but I cant seem to use those to move the entire level. The help files mention moving entire levels in world composition mode but not how, only the general set up process and benefits. Feel like I must be missing something obvious.
|
|
#
¿
Sep 8, 2021 00:37
|
|
|
SubNat posted:There's a couple ways to do it in BP (level blueprints can alter stuff like the transform of streaminglevels in it.) or C++ (There's a couple undocumented Fortnite world-streaming systems rattling around in the engine.) but it's way easier to just slap that plugin in and probably do what you want in a couple min. Man, that plugin doesn't work with UE5... I wish it did. Thanks though. Final scene size is going to be around 200m polys so I'm going all in on UE5 from the start. means lighting doesn't need to be baked or any UV's prepped too, which I think would really get in the way. I've had a poke around in bluebrints - I found load level instance, but honestly, I don't know what I'm supposed to be doing with any of this. If I play the blueprint, it deletes the rest of the levels I have loaded and only brings in the one that's in the blueprint. Do I need to rebuild all of the level hierarchy within blueprints? cubicle gangster fucked around with this message at 22:56 on Sep 8, 2021 |
|
#
¿
Sep 8, 2021 22:53
|
|
|
ImplicitAssembler posted:A fairly surefire way is to take it into Houdini, convert it to vdb's, vdbmerge and then convert it back to polys. You can do this with tyflow directly in max! It's got full vdb tools. And then max now has a remesher that many people argue is better than the one in zbrush. https://youtu.be/ktrIhk9NFT8 Both of those plus the new smart extrude tools have turned modeling in max into something really quite spectacular. It was already my favourite modeling tool but now it's just unbelievably flexible. And from what I hear, 2023 hasn't taken it's foot off the gas... I would also reccomend trying the regular tyflow boolean tools too - they produce a much better result than either of the built in max options. cubicle gangster fucked around with this message at 04:09 on Sep 12, 2021 |
|
#
¿
Sep 12, 2021 02:00
|
|
|
Felt like making something this weekend, but didn't want it to drag out. I recalled admiring the embers rising from fires in ghost of tsushima and that sparked the idea for a little sim & vibe excersize. Models from megascans. Phoenix for the fire, tyflow for the ash & embers. the ash does get caught up in the rising smoke but it's really hard to tell. there's also probably too much of it, and too small. oh well. https://www.instagram.com/p/CTwzewbrT5c/  I posted a long time ago about doing a tutorial - it's done! 90 minutes long across 18 chapters, focusing on working with filmed footage and retouching it like it's a still, but also covering techniques to work with 'final' rendered shots and retouch them to remove objects or add details. I fear finding an audience is going to be tough - it's very much aimed at people who don't know that this is possible, rather than someone trying to find a how to on something they know they want to do. Should be going live this week, hopefully within a couple of days.
|
|
#
¿
Sep 13, 2021 18:47
|
|
|
My big tutorial is done! https://i.imgur.com/cemt6sR.mp4 It covers painting out bits of footage and crane removal - but the technique is also useful on finished full CG shots and has a ton of other applications too. Part of the motivation to create this came from speaking to people about AE vs Fusion, and people who use AE on a regular basis simply not knowing why they’d ever use fusion (or nuke), or what node based compositing offers over AE. That is to say - if you’re thinking ‘I dont know why i’d need to know this’, maybe give the first 10 minutes a shot anyway - you’re the exact target audience. There is no rotoscoping or intense frame-by-frame animation required, and it can be done in the free version of fusion. It’s also relatively simple to do this in nuke if you’re familiar with all of it’s tools. Here is the link! https://www.youtube.com/watch?v=I4QzB7tT6zQ The technique goes back to something I bumbled my way through with the help of some people on the vray forum 12 years ago, and shortly afterwards got refined and moved over to fusion 6. (the first time I did this was all in max… rendering out entire sequences!) It seems like this approach is definitely underrepresented in tutorials, and so I felt it was a good subject to cover. I would not be where I am without endless free tutorials, other people sharing knowledge and having been on communities/forums my entire career. hopefully, this will help some others to do better, cleaner work. While it is long, it is intended to be extremely comprehensive. The technique is covered in the first 20 minutes, then the remainder of the video goes into a practical large scale application of this - and after i’m done, a bit of an overview of the pros and cons of various areas. It’s broken up into chapters for easy reference - ideally, after watching, you’ll understand the application and where it will be good to use, then when you come to pull it back up as a production reference ‘on the job’, however long later that is, you’ll be able to jump to a specific chapter. It’s not a day-to-day technique, but it turns what is an absolute pain in the rear end job using AE into something that is easy, flexible, and quite enjoyable to work on. If you get 20 minutes in and think this is something you could use, I would recommend watching it in full before trying to ‘follow along’ on a live project - there is some insight towards the end when i’m covering the finished document about what worked well and what I had to fight with a little. I tried to cover every single application and thing you might encounter in the same project, but it did cause a few challenges that could have been avoided (but then, i wouldnt have been able to show working around them...) I sincerely hope people are able to enjoy or get something meaningful from this. I imagine there are more ways this can be refined further - if it does spark an idea or you have a way to take this even further, i’d love to hear about it. You may also notice it’s being hosted on a DBOX ‘Dialog’ channel... There’s a good chance there will be more of this - covering less obvious & underutilized production workflow tutorials. cubicle gangster fucked around with this message at 00:13 on Sep 26, 2021 |
|
#
¿
Sep 25, 2021 17:03
|
|
|
Unless this is what you need to rig - https://m.youtube.com/watch?v=SfExejqdr5w
|
|
#
¿
Sep 29, 2021 14:28
|
|
|
ImplicitAssembler posted:Right, but you'll still get most of the default behaviour from rig building tools. Admittedly, I haven't kept up. I switched from rigging to FX some 15 years ago Yeah, this is definitely more of a logistics & planning challenge, I just saw it recently again and thought it was worth sharing! I'm pretty happy with how that tutorial is going over - I only shared it to my personal facebook, here, and a couple of other forums. After 3 days it's up to 1100 views with a huge number coming from whatsapp, gmail & slack, which means people are sharing it with each other. Viewer retention is surprisingly good too - 40% of the people who make it past the first minute stick around for at least 20. Analytics are great!
|
|
#
¿
Sep 29, 2021 19:32
|
|
|
You're already fully in a motion designer career? stick with c4d - you might want to be able to get freelance work from someone who needs an extra pair of hands. C4d is loaded with mograph specific tools too. Blender is great and everything does translate to other software, but it's best for someone dipping their feet in without a career path in mind already. Fantastic tool for hobby/personal work, and worth having alongside your main. I dip into it once every few weeks for the things it handles better than max, but my industry is max and any new hires are essentially a junior until they become proficient - ability to hit the ground running does affect people's ability to negotiate a salary.
|
|
#
¿
Oct 2, 2021 07:35
|
|
|
Based on the questions you've asked, I don't think there's a single piece of 3d software (excluding zbrush, Houdini & 3dcoat) that can't easily do what you want. Blender, max & SketchUp would be the obvious choices. Rhino & fusion runners up. cubicle gangster fucked around with this message at 06:14 on Oct 18, 2021 |
|
#
¿
Oct 18, 2021 06:12
|
|
|
Id model it with as many overlapping objects as you need, then when it's done use tyflow to turn it into a vdb with an extremely high density, watertight/single volume, 20m poly+ object. That will give you good control over the blending between intersecting objects too, then use the new max remeshers and auto retopology to bring it back down to a manageable polycount before printing. Max is better at retopo than zbrush since 2022. cubicle gangster fucked around with this message at 03:50 on Nov 11, 2021 |
|
#
¿
Nov 11, 2021 03:47
|
|
|
Use a proper camera matching tool. Im not sure how knowing the field of view can tell you anything about the camera height. What did you measure if its above the sea too? I googled 'blender camera match script' and it seems like there are a lot of options. There's also a process to do it manually but it's kind of long winded and I use max. If you're willing to share the photo I can give some quick pointers that might make it easier, probably involving editing the image before you start. Don't forget drone photo metadata usually has coordinates embedded - they're not perfect but you'll be within 15ft and can manually refine from there. cubicle gangster fucked around with this message at 16:04 on Nov 26, 2021 |
|
#
¿
Nov 26, 2021 15:49
|
|
|
A project which has consumed a phenomenal amount of my time over the last 18 months is finally launching. It's a four seasons private residences in Austin, overlooking the lake. Might be on the most incredible real estate site i've seen in my entire career. I was on site again a few weeks ago with a dual 5ds setup getting some stereoscopic landscape photography from a tower that's been built - getting up at 5am for dawn and watching the sunset every night up there was pretty special. I've been leading most of the images in Miami with everyone here, but we've had a lot of people at dbox contribute to and work on them, including some of the guys in our new Ukraine studio.    https://www.fourseasons.com/residences/private_residences/lake-austin/ We've done around 80 images in total (so far), but only the ones at this link are currently public. One reason we've been on this for so long, and done so many images - Matthew bannister, CEO of DBOX, is the lead design architect on the project. His first built work will be a $1.4bn four seasons... not a bad portfolio opener! but basically it's been really exciting for us to go beyond a service of making images, we've been super close to this project every step of the way. The client has been amazing, an absolute joy to work with. he's the perfect client, and that's likely because he doesnt have any real estate experience - he made his money in tech. Fun fact - richard garriott of ultima fame lives right next door, with his mini recreation of Shakespeare's theater and a private renaissance fair type setup along the riverbank. That's all hidden in the trees in the lower left corner of the aerial image. cubicle gangster fucked around with this message at 01:32 on Dec 2, 2021 |
|
#
¿
Dec 2, 2021 00:14
|
|
|
Thanks! Hopefully, we'll have more to share soon. The fitness center/orangerie/spa is astonishing, and then the penthouse floorplan layouts are flawless. Lots of really spectacular spaces, makes the CG side a lot easier! Sales have far exceeded everyone's expectations so there's been no need to make much of it public yet, but we will be showing it all eventually.
|
|
#
¿
Dec 2, 2021 17:39
|
|
|
I thought maxon was part of Adobe already. The newest versions of AE also install a full standalone version of cinema 4d that pushes you to add it to your Adobe account as an optional extra.
|
|
#
¿
Dec 14, 2021 20:47
|
|
|
the_lion posted:Nah, C4d in AE is just a lite version, not the full thing. A standalone copy of cinema4d appeared in my start menu, it brought the full package with it. I haven't used it because I'm not registered but it seems like it's potentially the full install.
|
|
#
¿
Dec 15, 2021 02:46
|
|
|
I'm going to have a play around with it this weekend! The licensing is bizarre. For every hour you use it, it extends the license 14 days? So if you stop using it, you can't use it anymore? That's the best I can parse it and it doesn't make a lick of sense.
|
|
#
¿
Jan 21, 2022 06:30
|
|
|
At DBOX we tack on $150/s to cover all rendering of a film. That's covers general hardware depreciation/investment, cloud rendering if required, software licenses & storage. You probably wouldn't be able to go as high, but you should absolutely get paid for rendering full CG animation separate to the work in putting scenes together. If you're solo, you can do some quick math to figure out your number. How much space does each jobs render output take up on a HD and Dropbox backup, what % is that of how often you pay it, what hardware do you need divided by the number of projects you can use it on, what kind of time does it take to deal with the hardware/storage etc. Say you do 8 projects a year at 20s each, use free software, you want 3x 3090's and to replace them every 2 years, and each job takes up a tb - you end up with a number around $500 per project, or $25/s cubicle gangster fucked around with this message at 21:19 on Mar 8, 2022 |
|
#
¿
Mar 8, 2022 21:07
|
|
|
Watch more movies? Your cameras are all insanely wide lenses, and youve got things like a handheld style wobble on a motion that would be on a crane shot. Yes it's cg and you shouldn't copy real cameras forever, but I think you'd do well to spend a while sticking with realism and learning the rules before you throw them all out. If a camera is handheld, think of the limitations of that and stick to it. If it's high and a crane movement, keep that crane style, and study what that should look like. If it's in a tight space, you're going to use cuts between fixed cameras.
|
|
#
¿
Apr 30, 2022 20:09
|
|
|
Songbearer posted:I'm not a film buff of any description which makes it very odd that I enjoy animating so much, I just don't have the attention span for films v For what it's worth, a movie is only a 2hr investment - start watching stuff that doesn't interest you based on it having a reputation as something that is shot very well and just stare at it with a critical eye. A miss is not a lot of time lost, and a hit may be something that you find very special. I didn't like movies until I started to become more critical of them, and there are some very, very good ones out there.
|
|
#
¿
May 2, 2022 04:28
|
|
v that said I'll start really looking at the things I'm watching and see how they do things.
|
BonoMan posted:Speaking of unwrapping - I'm all in on Rizom UV. Holy crap I love it so much. I'm in C4D and have the bridge installed and it's just so freakin' easy. Rizom is great! we dont have to unwrap things much, but a handful of times a year some nightmare situation comes up and it pays for itself in an instant.
|
|
#
¿
May 31, 2022 17:48
|
|
|
bop bop perano posted:I was reading this article on how this guy made a car from the game Disco Elysium and also downloaded what I think is the same model and it was a bit depressing because I feel like I would barely know where to start, attempting to make something like this, how they made it. he's got years of experience and is modeling in a fairly complex way specifically for printing, but also to a standard that gets a good game/vfx model out at the end - it's pretty reasonable that you wouldn't be able to start at 0 and jump straight into this. It's an article about the process aimed at people who have a few years of experience but want some insight into new approaches, not beginners. You will need to learn subdivision modeling, just start there (not trying to remodel this particular car yet) and give it a couple of months of practice. You will not learn 3d software by trying to figure it out as you go, it's not designed to be used like that. put aside a couple hours a few times a week to follow tutorials, and after a month or two then try remodeling the car, piece by piece - and when you get stuck, look for specific tutorials about how to achieve particular shapes. If the only reason you'd ever want to learn 3d software is to do this specific thing, you may be in for a rough time. if this has inspired you to learn 3d software and you want to make all sorts of things, you'll be able to enjoy the process. There are an insane number of detailed modeling / texturing / lighting etc tutorials for blender out there on youtube.
|
|
#
¿
Jun 7, 2022 19:02
|
|
|
A while ago someone in here recommended a remote desktop solution that was designed / positioned for the video game industry - ultra low latency so you could play games through it was it's big feature. Anyone remember what it was called? We've got someone having a bad time on splashtop & windows remote desktop and this would be good to try.
|
|
#
¿
Jun 8, 2022 18:19
|
|
|
Awesome, thanks guys! It's not their internet - they've got 900 down / 300 up. Physical distance is a bit more than usual and I guess splashtop is really sensitive to that. it's only for 6 months too.
|
|
#
¿
Jun 8, 2022 23:10
|
|
|
Smashing it as per usual. Great to see your new work!
|
|
#
¿
Sep 4, 2022 07:09
|
|
|
This little project has consumed me entirely for the last 4 weeks, but now it's done. https://vimeo.com/748129523/fa84008a3e I shot all the footage on an inspire 2 with a coworker last year, we've been sat on it since. 4 weeks ago we had a shortlist of 30 ungraded shots and an absolute mess of revit models with no details. I was in charge of the footage, tracking, lighting & integration, and 4 people on my team worked on the model - cleaning it up, adding details, getting the trees, roads and paths in etc. It's been a pretty heavy lift! 
|
|
#
¿
Sep 9, 2022 19:29
|
|
|
BonoMan posted:This is insane. Excellent work. Thanks man  In a couple months we should be rolling into a full 4 minute film of this project covering every part of it, looking forward to getting stuck into that!
|
|
#
¿
Sep 17, 2022 19:06
|
|
|
|
| # ¿ Apr 25, 2024 11:23 |
|
|
Prolonged Panorama posted:Very nice. How do you handle the transition from plate to CG, especially with trees? Paint out a few tree-widths away from what you need to replace on the plate, and use the hard alphas of the corresponding replacement trees in CG? Or do you have a CG forest that extends further than you need and you transition with a feathered mask? Or something else? Hopefully the question makes sense. Pretty much both of those! The base render had trees covering the entire frame and hillside. Broke it out so i had my base CG trees color corrected as close as possible to the plate trees - this left me with a set that matched in color, but they had really fine 'cg' like details and the contrast was off, but it worked close to the architecture. Then I broke that out into another set where I blurred the plate a ton, reduced the contrast, took the raw light pass from CG and multiplied that against it - gave a set that had exact color from the plate (and all the subtleties that didnt come through with CG trees), but cg details. Then I took the raw light pass, blurred it, and did the same again with a less blurred plate to feel softer and even closer to the footage. That gave me 3 layers of transition to go from footage to CG - which was pretty easy to use in fusion, I brought a simplified terrain model, looked at it from a top down and drew a big soft mask which was then rendered from the animation camera. using different contrast/blurring levels of that mask multiplied against the tree multimatte gave all the layered masks blending layers to transition nicely. The second shot skimming across the water, mostof this went out of he window. it's got some issues and to do it properly needed more work than i realised - you can see the plate trees sliding underneath the CG light/shadows in a lot of places. It wasnt perfect but I got it to a place where it's hard to notice the issues on first watch, and thats good enough for the budget/time we had.
|
|
#
¿
Oct 5, 2022 17:23
|
|