Home General Discussion

photogrammetry will change the game?(industry)

polycounter lvl 7
Offline / Send Message
ffm32206600 polycounter lvl 7
hello
as u know these days photogrammetry is being very common in most cases,but u think it will change the pipe line industries very much?
would be environment artist go way from most of jobs?
my opinion is it can be used only for very little percentage of models specially organic models like as rocks ... but for example for modeling a car i dont think photogrammetry would be useful enough...
generally what do u think about this manner?im interested to know your ideaas...
tnx

Replies

  • Eric Chadwick
  • poopipe
    Offline / Send Message
    poopipe grand marshal polycounter
    Imo as a studio there's no point in doing it internally unless you're talking about huge budgets and very specific source material (eg battlefront)
     You can buy very good quality data off megascans  for very little money compared to the cost of capturing it yourself and there's plenty of tools floating around that allow you to muck around with it.

    I disagree about how useful it is.  Just having something to draw around can save days and days of work, never mind all the other data you can pull from it
  • Zi0
    Offline / Send Message
    Zi0 polycounter
    Its pretty common in AAA developments these days, a lot of studios have been using it for half a decade now.
  • ZacD
    Offline / Send Message
    ZacD ngon master
    Personally, I think the future and when photogrammetry will really take off is when AI/procedural tools will let artists remix photogrammetry data into new assets.
  • thomasp
    Offline / Send Message
    thomasp hero character
    Photogrammetry at least for characters has been around for a long time. I encountered it in 2006 at the start of the PS3 generation, far from being an early adopter. It appears to have bumped up the quality expectations across the board, not killed off jobs. I assume its the same for environments.
  • poopipe
    Offline / Send Message
    poopipe grand marshal polycounter
    ZacD said:
    Personally, I think the future and when photogrammetry will really take off is when AI/procedural tools will let artists remix photogrammetry data into new assets.

    That's been a thing since day1 
  • DebendraRoy
    Hi, I also do wonder about how 3d scanning hasn't made modelers/texturing artists dealing with realistic humans redundant. What are the exact reasons that there is still a need for sculptors while creating digital doubles? Thanks.
  • pior
    Offline / Send Message
    pior grand marshal polycounter
    "What are the exact reasons..."

    3d scanning only does what 3d scanning does - nothing more, nothing less. To the layman it will always seem like surface details sum up the whole thing. In practice, these things are probably the least important factors contributing to a convincing character (or rather : they are the easiest parts to get right).

    With all the knowledge, skills and tech available, even the most advanced studio in the world will only be able to make mediocre content if things are not carefully crafted. Here's an example of the best output an AAA studio could do in probably a day of so worth of time - quickly fixing up the scans and probably setting up the head to a generic rig without much tuning. It's pretty much awful.

    at 4mins30, then 6mins12 for the result.
    https://youtu.be/KlMpN26ts7s?t=372

    The best way to get a grasp of the reality of these constraints is simply to make a character head oneself, from start to finish, with facial expressions - even in a cartoony style. It's a tremendous amount of work with or without 3d scanning.

    And then of course there's the LA Noire tech, but that a completely different story.


  • sacboi
    Offline / Send Message
    sacboi high dynamic range
    my opinion is it can be used only for very little percentage of models specially organic models like as rocks ... but for example for modeling a car i dont think photogrammetry would be useful enough...
    Quite the contrary... 

    The author's scanning setup:
    (Note a few years old, may've upgraded since then)



    OP, yes of course entirely useful but I'm afraid for the foreseeable future at least, learning conventional fundamentals toward that which ones envisioned goal mastering, is definitely the most worthwhile course of action to embark upon.  
  • DebendraRoy
    @Pior Thanks for replying. & thanks for the timestamps... But I see Coco, I watch the whole thing..  :)

    I'm still not getting it though. Won't scanning give you the likeness faster than any 3d artist could? & then it's just a question of retopologizing the scan? What am I missing here?

    Or is cleanup much more tedious and time consuming?
  • pior
    Offline / Send Message
    pior grand marshal polycounter
    A likeness sculpt (or scan) is the easiest and most straightforward part of the process. Achieving it through scans takes a few days since it needs cleanup. Achieving it through sculpt doesn't take much longer - and there are many, many people able to do that, with varying amount of quality but it's not a rare skillset. The part that takes a tremendous amout of time to look good is what comes after (the part you describe as "just retopologizing"). As said : go ahead and build a head model (doesn't need to be realistic), and take it all the way through to a final convincing asset able to strike a solid range of expressions and looking good from all angles, working well in an engine. This takes an great amount of work and skills - scans or not.

    Your misconception is to assume that the scan holds most of the "stuff" that makes the head work. It absolutely doesn't :)

    Of course anyone can do a cheap scan at home, decimate it, slap some automated UVs on it and stick it into Unreal - that's easy and takes a couple hours from start to finish. But that sort of stuff is nowhere near an actual production "double".
  • DebendraRoy
    Ah, okay! I get your point . Thanks for your patience and the detailed reply, :) . Cheers. 
  • gnoop
    Offline / Send Message
    gnoop sublime tool
    In my experience  photogrammetry while  taking time  still way, way quicker than  sculpting and  especially painting something  either procedural or manual at a same level of visual realism.            It just needs  a  quality shooting  job that itself require money,  shooting trips,   people, equipment etc.
    Games doesn't do it much because   their goal is  not ultimate realism, rather something decently looking  for a cheap price. 

    What surprise me more is why we don't see  much for sale.   I can't stand  Megascan 2x2 meters textures repeating like crazy  so have do do it my own  10 meters scans.        Neither we have any convenient soft that could easily  bake hi res details , make variations  or just convenient re-compositing.      Even Artomatix  didn't fall in that category when I used it last year.

  • Blaizer
    Offline / Send Message
    Blaizer polycounter
    It was said 3D would replace traditional 2D animation, and look...

    3D Scans and photogrammetry are good for games with "realistic" people and environments. Did you watch the unreal engine ps5 demo? the female character doesn't match/fit in the whole. It looks weird!
  • Alex_J
    Offline / Send Message
    Alex_J grand marshal polycounter
     I also question if the money involved in chasing more realism via scans or otherwise is really worth it.

    How much is it just professionals doing their thing and always upping the ante versus what the market actually demands?

    I guess I'm drifting out of the market as a consumer but honestly I am still perfectly happy with Dreamcast graphics. I still think the photos wrapped around sphere heads in original Shenmue have great charm and I can get immersed in that shit.

    I imagine there is probably millions of people who buy the yearly COD just because it has slightly improved graphics from the previous. But at what point do you hit diminishing returns? And if TLOU2 didn't update anything about the game other than the story since the previous, would it sell any worse? I wouldn't have cared at all.

    I think there is probably a lot of money wasted in keeping up with the Jones's. I guess you got to know the full picture but it seems to me like when you got three different artist making 15 textures and traveling around the world just to make one 3d character come to life, it's like, how does that ROI?

  • sprunghunt
    Offline / Send Message
    sprunghunt polycounter
    Alex_J said:
     I also question if the money involved in chasing more realism via scans or otherwise is really worth it.

    How much is it just professionals doing their thing and always upping the ante versus what the market actually demands?

    I guess I'm drifting out of the market as a consumer but honestly I am still perfectly happy with Dreamcast graphics. I still think the photos wrapped around sphere heads in original Shenmue have great charm and I can get immersed in that shit.

    I imagine there is probably millions of people who buy the yearly COD just because it has slightly improved graphics from the previous. But at what point do you hit diminishing returns? And if TLOU2 didn't update anything about the game other than the story since the previous, would it sell any worse? I wouldn't have cared at all.

    I think there is probably a lot of money wasted in keeping up with the Jones's. I guess you got to know the full picture but it seems to me like when you got three different artist making 15 textures and traveling around the world just to make one 3d character come to life, it's like, how does that ROI?

    For characters I can't see why you'd need to travel - mostly you'd use scans of fabric, metal, and leather. The same sort of thing they do in VFX. 

    But for environments there have been several games where the environments use 3d scans extensively. For a AAA game it's not a big expense to send someone to another country with a rig setup for photoscanning compared to having a team of dedicated artists or outsourcers create the assets. 
  • Alex_J
    Offline / Send Message
    Alex_J grand marshal polycounter
    I guess the expense is worth it because they're doing it.

    Trying to make something from nothing though I really pinch pennies. Really hard to imagine the kind of money put into the big games.
  • sprunghunt
    Offline / Send Message
    sprunghunt polycounter
    Alex_J said:
    I guess the expense is worth it because they're doing it.

    Trying to make something from nothing though I really pinch pennies. Really hard to imagine the kind of money put into the big games.
    If you're doing things on the cheap side photogrammetry can help if you buy assets from megascans or other places, like the unity asset store, and then you don't need to make them yourself. Then you get a detailed result for very little cost.
  • Ruz
    Offline / Send Message
    Ruz polycount lvl 666
    I expected to see a photo of a gramophone in this thread - sad face
  • another caveman
    Offline / Send Message
    another caveman greentooth
    Photogrammetry got an important spot in my VR workflow.



    Things like this are a huge gain of time, but also as simply as using google maps model/texture to draft things out etc.

    ps: photogrammetry also is great for non realistic etc.
    I find it to open the door to more artstyles in fact.



    Download meshroom, grab a camera, craft any shit you want out of whatever materials.., it's became so easy to scan things and pop them in your game

    Use a piece of cloth to be your ground etc...


    Also there was this great use of PG I cannot find again the video.

    They had kids play in a sandbox, make a track / level, and then hit 'play' it would scan the sand box, and have them race on it :)

    anyway it's one more tool to my set !
  • Larry
    Offline / Send Message
    Larry interpolator
    Blaizer said:
    It was said 3D would replace traditional 2D animation, and look...

    3D Scans and photogrammetry are good for games with "realistic" people and environments. Did you watch the unreal engine ps5 demo? the female character doesn't match/fit in the whole. It looks weird!
    Unless you scan a person. I think it was deliberately done to showcase a bigger contrast
  • another caveman
    Offline / Send Message
    another caveman greentooth
    the iphone and lidar combination or skydio and lidar combination look very interesting
Sign In or Register to comment.