What makes the rendered Daz human images still distinguishable from the photos of real human?

2»

Comments

  • CricketCricket Posts: 460

    I think body language plays a part as well. DAZ figures tend to look stiff and not relaxed - no shoulder drops or poor posture.

  • McGyverMcGyver Posts: 7,053
    edited August 2023

    I'm guessing if it's being shown somewhere related to 3D rendering, or at DAZ, like in a gallery or post, I already know it's a DAZ rendered virtual person, so then other things instantly become apparent... but I'm probably less likely to notice (a really well done render)  if it's just a cold viewing of a random image being used in an advertisement or a book cover... 

    There are all the dead giveaways people have mentioned beforehand, like hair, slightly unnatural posing, too perfect, etc... but when someone does good postwork, it bridges that uncanny valley... but nowadays most times it's usually context... I'm expecting it being CGI so I look for telltale signs.

    For example I recently started noticing a lot of clickbait ads that used to use real (human) buxom young females that have nothing to do with the subject (like ads titled "The Government is GIVING AWAY FREE solar panels NOW!" or "What your DOCTOR doesn't want you to know about SQUID JUICE!", but for some reason it's a thumbnail of a busty young lady in a tight T-shirt in a random location like a courtroom or next to an industrial air conditioner)... these ads which generally used to use actual people are starting to use AI generated images that seem to be cobbled together out of real images and CGI... 

    Now that I'm aware I tend to notice it more... it originally started with badly made images of houses with wavy distorted solar panels on the roofs and some odd car or landscape images, which were so bad you couldn't not notice, but now its people too... and even though they are a lot better than the old ads, it's stuff like not showing hands or oddly blurred hair that are the telltales.

    I suspect if you hate clickbait you ignore it, but if one is easily distracted by extremely round objects tightly confined by cotton fabric, they are more likely to be distracted by the virtual representations of temperature activated central surface irregularities and miss it... and probably click on the ad hoping for more revelations of such irregularities... possibly even going as far as to purchase a gallon of refined squid juice infused with magnetic silver ions, in hopes that one may attract similar geometric anomalies to their vicinity.

    Rats... I forgot what I was going on about because now I want to go out and buy cantaloupes and honeydew melons for some reason... 

    Whatever... it's often context that makes me notice unreality, otherwise it's easy to gloss over if one isn't overly invested in the image or examining it too closely.

    Well... it's off to the melonery... melon-o-porium... place with melons... Supermarket!... Wow, why was that so hard to remember the name for?

    Post edited by McGyver on
  • Nyx, you're completely right on that. The G9 base poses and the V9 standing poses are quite good in that respect.

    With poses it can be hard to see what is wrong, even though you get an impression/feeling it's not right. When I generate my poses I try to act them out. If it feels wrong, I change it until it feels comfortable. I get the impression not a lot of people posting images do that. The skill after trying out the pose, then, is to replicate the exact position your body has actually acted out as opposed to the position you think your body has acted (I'm gettting to an age where the two are not automatically the same).

    Regards,

    Richard

  • AgitatedRiotAgitatedRiot Posts: 4,437

    I would say that it's because they are not real.wink

  • oddboboddbob Posts: 396

    Blue skin, pointy ears or tentacles are a bit of a clue as well.

  • WendyLuvsCatzWendyLuvsCatz Posts: 38,229

    some of the furry conventions have cosplayers that are replicated pretty convincingly in 3D though

  • FirstBastionFirstBastion Posts: 7,762

    This looks like a real world snapshot.

  • LauritaLaurita Posts: 222

    Sorry, at best this looks like a real world snap shot of a painted miniature figure. The cloth texture looks like tarpaulin, the straps of the attached polches don cut intio the pants, the belt is just the same and even the grass doesn't look natural. Although I can not fingerpoint why the grass looks off. Oh and the shiny and brand new leather texture of the pouches just isn't fitting to the weathered appearance of the clothes and boots.

  • lilweeplilweep Posts: 2,490

    FirstBastion said:

    This looks like a real world snapshot.

    Figure and environment are really nice at first glance, but outfit begins to look like a thin Marvelous Designer outfit (which is obviously was). Tell-tale signs of deformations from rigging also give uncanny valley vibes.

    I think if Male Imagery used some photoscan assets of military garb rather than his own (thin MD garment) he could achieve something very good.  Becasue military outfits are in demand, there are quite good ones on various general asset stores you can get.

    Male Imagery is one of the few people whose daz figures actually look very convcingly real in certain renders.  Perhaps because theyre in Blender, idk

  • jd641jd641 Posts: 459

    richardandtracy said:

    Nyx, you're completely right on that. The G9 base poses and the V9 standing poses are quite good in that respect.

    With poses it can be hard to see what is wrong, even though you get an impression/feeling it's not right. When I generate my poses I try to act them out. If it feels wrong, I change it until it feels comfortable. I get the impression not a lot of people posting images do that. The skill after trying out the pose, then, is to replicate the exact position your body has actually acted out as opposed to the position you think your body has acted (I'm gettting to an age where the two are not automatically the same).

    Regards,

    Richard

    No one gets posing right, not even the best of us. The only way something looks natural is if it's been taken from a motion capture file, even then it may not be absolutely spot on due to the figures design. One of the biggest offenders to me is when a model is posed laying on their side/hip with the arm supporting the torso... it's a typical pinup pose that's done over and over again but man everyone gets it wrong. I literally feel pain in my side, knees and my shoulders when I see these type of images hahaha.

    As far as anything else goes, human eyes easily see what's wrong, daz figures (more typical of the females) have a percentage of cartoony-ness to them that immediately alerts me to "something isn't right with this". Until that changes at least for me, it's still going to be very easy to tell the difference between what's a photo of a person and a render even if the skin/eye/hair looked photoreal.

Sign In or Register to comment.