3d Scan Human Head with Reality Capture

what's going on so we have scanned tree trunks we've scanned rocks in grand canyon now obviously the next step up from that is scanning live ducks so we are here at the park see some ducks hanging around over there somewhere so i'm going to try and see if i can photo scan the ducks because i know people are getting tired of rocks so i got your ducks right here and hopefully they don't bite me no i'm just playing um i'm going to try and find somebody in the park now i went in the coolest part of the day here which is right now it's 109 degrees which is the coolest part so i'm out here i'm hoping there's somebody out here just kind of hanging out and i'm gonna try and see if they'll agree for me to take a picture of them for about an hour you know because photogrammetry takes a long time so yeah hi do you have like 45 minutes um i'm taking a photo like photogrammetry stuff do you mind if you like my subject today there's like nobody else here but you okay is that okay i mean you're already like sitting there so um do you mind putting this bald cap on because it's it's gonna look really weird with hair flying away and stuff it's really hard to do so if you don't mind that'll be super cool my name is jay by the way kristen nice to meet you so um let me go turn this off and i'm gonna go set up my behind the scene camera which is gonna be my cell phone but yeah so like i mentioned in my previous videos when it comes to photogrammetry you want to take pictures uh during overcast days now unfortunately this park right here doesn't open up until like eight to nine am and by that time the sun is actually pretty high already at this point i did put her under a tree so she gets a little bit of a shade but as you can see in this image right now her left side actually her right side of her face is in the shadow just a little bit and unfortunately there's there wasn't really anything i can do just because of that time of day now like i said in the video this was probably the coolest part of the day as far as the park being open so i just really had to make two but whenever you're planning your photogrammetry go ahead and plan for overcast days or when it's like cloudy so that you'll get an evenly lit subject okay so here we are in reality capture software and what i'm going to do now is try to recreate the photos we took of kristin in cg or 3d okay so quick start i'm going to click on that i'm going to change my layout to one plus one plus one is equals three okay she didn't know that man i'm feeling smart today so i'm gonna go ahead and go to the folder and i already created a folder here titled a as you can see here we have 101 images now i wish i could have taken more but i thought that it was gonna be enough but that's okay uh we're gonna try and work with this right here so we have the images as you can see i pretty much just went around christian here and like i said in the previous voiceover her face is in the shadow just a little bit i mean by not a lot just just a little bit and when i was taking her picture it didn't look you know too bad let's go ahead and take a look at this now i already removed i think i had one blurry picture and that was it which is pretty awesome so you can see i shot it at f35 so what i'm going to do is go to the alignment right here and i'm going to go to the settings and i changed this because this is 10 000 by default and this is like 40 000 so i changed it to 100 000 and i changed the image overlap to low which let me change that to medium so you're gonna have to play around with this the settings i'm showing right now might not actually be the same that's going to work for your photogrammetry okay so let's align images and see how well the software aligns this image here okay and it looks like it created two components one has 70 pictures and one has 17 pictures so i'm going to take a look at the 17 one real quick and let's take a look i'm going to change this to my 2d view alright so that's going to be tough those back ones those are going to be really hard because there's not a lot of thing that kind of sticks out so what i'm going to do is delete this and then try to realign it again so i'm gonna delete all this and see if i can realign it one more time so i'm gonna change this to low and see if this will yield a better result for us sometimes you just kind of have to play around with the settings because sometimes i don't even like change any setting and i just realign the images and it just does a better job but basically you want as um minimum of components as you can meaning like 101 out of 101 so 17 is quite a bit because you know it's gonna be a lot of realigning manually so let's see if we can get a better result by just changing that to low and then realigning it again all right so i'm liking this number here because we got 89 chunk two and then five so again we're going to check out the 2d right here and let's look at the component zero let's see what happened which one probably the same thing okay so that one is the right side of her face which is good because i can easily align that okay and you know scanning human obviously this is my very first straight human photogrammetry right her face is pretty smooth for the most part right because it's not going to have a lot of you know like a rock like a rock has so many edges rough edges and it's easier to kind of like create or recreate in 3d but her face is pretty smooth for the most part okay and thankfully she didn't actually wear too much makeup because if she did then i would have been really screwed because her face would have been just really smooth with no surface so i'm gonna take that as a five component is good i'm happy with that and this two images right here i'm gonna open up okay so again the back right and obviously reality capture is just having issues as far as aligning this with the rest of the image but what i'm gonna do is save this for now okay so we're gonna take care of the big chunk first of the five now if you have any better tips on how to do this let me know but this is kind of like how i learned it from the documentation so i'm going to go to component 2 and like i said we're going to take care of the five images first so the the right side of her face so i'm going to go to component 2 which is where our majority of the chunk or images oh look at that that's behind the scenes camera right there buddy so i'm gonna go on the other side of her face i'm gonna look for it somewhere if i can find it okay so here's one 660. now what i'm going to do is actually do a one plus two plus two and i'm going to change this to a 2d view change it to a 2d and this is a 2d and i'm going to drag that 660 right here and as you can see 660. i'm going to find another one that kind of looks like that 635 yes i'm going to drag that there let's look at 658 yes i'm going to drag that as well and let's see that's kind of harder let's do a 632 i'm gonna pick all these images pretty much showing the same exact features of what's missing so if i look at this picture right here and drag and drop it this image was not able to connect to the rest of the chunks so basically what we're doing is we're connecting these five pictures so that it's gonna belong right here it's gonna be added to the component two so we're gonna try and do merge component zero with component two that's the goal so uh let's go to 660 again because i think that's what i just pulled out right oh man okay let's just pick another one that has the same features doesn't have to be the same exact one so 632 i already have 632 633 is really close 661 do i have a 661 okay that's close enough now you can kind of see why i asked her to put the um bald cap on because if i didn't her hair would be everywhere right now i mean you can see there's a little bit of hair right here um but as much as you can you know get a bald cap and just control the hair you know because it's gonna help you in the long run here so what are we gonna do now we're gonna do is actually create control points on all of these images in component two so click right here i'm gonna pick these three right here because they're pretty prominent on all the images as you can see i can see all of them here i can see it here and i can also see it in the image up here from component zero so that's why we're choosing this so i'm gonna go right here that's gonna be point zero and i'm gonna tell it you know it's actually on this one and then down here this one and then down here this one so that works out and then what we're gonna do is go back to the first image create a point one now okay point one is right there and then point one is right there point one is right there again create another one two two two okay so now we connected all four of these pretty much okay this is the same exact uh position as these if you're familiar with tracking say like after effects you know what this is you know you pick out the best contrast contrasted area on her face which like i said is good that she didn't wear a lot of makeup because it helps us out as far as you know getting that tracking whenever the software is trying to figure out what her face should look like okay so now that we have those done i am now going to minimize this and i'm going to uncheck control points and i'm going to start dragging the component 0 images so here's one and as you can see we can see those three dots and let me check 697 again same thing and let's do 696 again same exact thing all right so i left this one right here because i'm actually gonna use this reference so what i'm gonna do is click on point zero press ctrl and i'm gonna drag and drop that to the component zero images i'm gonna click on point zero right here okay then i'm gonna go control point zero right there and then point one same thing click control drag it to that reference and drag it to our reference then we got point two boom oh we're missing point one but it's okay there's point two and i'm gonna drag point one up again there you go so point one okay and then we're gonna grab two control two right so now we just told the software hey for these three images use this they should be pasted right here okay so now what we're going to do is just press uh let me see let me click on point two again so that will disappear click on shouldn't have anything on my cursor here align images again okay so now it created another component now we have 94 of 101 which i'm actually pretty happy about so if i go to component 2 we had 89 now we just added these three images to this component 0 which is good because i'm only missing a couple of images so i'm going to double click that and i'm going to i'm already in 3ds so i'm going to go go around i think i'm what i'm going to do now is just make sure that i'm setting their reconstruction region oh control z on that one [Music] make sure that her entire head is covered just like so i'm not too worried about the ground plane here because it's really just the neck in the head that i really want so i'm gonna save this now so from here i'm gonna change my layout to just one plus one and i'm gonna go back to help right here i absolutely love this this is such a good idea so double click right here i'm just going to move around so if you want to get rid of those little annoying cameras if you go to scene you can just shrink those alright so the next step is actually going to be computing this to become a 3d model so to do that let's go back to workflow i'm going to go calculate model i'm going to choose normal quality all right so it looks like it calculated our model here and as you can see right here the right side of our face is a little bit messed up especially around this area and obviously i didn't have a ladder so the top part is screwed but that's okay i mean technically you just really need her face right here now there's a lot of ways you can clean this up you could probably take this in z brush which we're going to try and see but right now what i really want to do is just get this thing textured up to see what it will look like because i'm pretty excited to see how well this worked for us so what we're going to do is just do a lot of automated cleanup tools here so i'm going to go to reconstruction and i'm going to click clean the model all right that's good and i'm going to do smoothing tool as well and i'm just gonna smooth and leave everything by the fault and what we're gonna do next i'm gonna go ahead and get rid of these right here i'm going to go ahead and unwrap this thing here okay so i'm going to go to unwrap settings right here we're going to use everything like we've done before 10 10 2 8k texture unwrap so as you can see we're at model 3 now because we were cleaning it up and doing a smoothing tool so it's creating a different version each time you do that so next we're going to do is change this to 2d and we're going to go to images and look for a good color corrected image i don't know why that looks kind of pink but i'm gonna go to selected input right here and i think i already yeah i already did this one so i'm gonna go enable right here and then i'm going to correct colors right let's make sure everything is good here so let's go back i'm going to minimize this double click our model 3 make sure everything is good let's take a look at the settings for this let's go check out check out the texturing we're getting 120 texture quality which i'm pretty happy with that's good so it's going to create three textures as well looks good and now what we're going to do is uh texture projection so we're going to take this model 3 and we're going to take the texture information from model one okay so to do that we're just going to go to reconstruction again texture projection and i can minimize this so it doesn't confuse a lot of people selected input as well so right here we project model texture i'm gonna say the source model is gonna be model one and we're gonna go with the result as model three so we're gonna take the texture information the normal map from this and we project it to model 3 which what we want and i'm going to have normal group project entered on as enabled all right so what we're going to do next is just press texture okay so it looks like it is done texturing and we now see our final textured model and yes you can definitely see the difference between the light side and the shadow side but without any actual manual cleanup this is actually pretty darn good all right so that's pretty much it for this video now i'm gonna have to find a way to clean this up in another software because i don't have zbrush it's not a bad scan i don't think i mean obviously the like i said the shadow and light area could have been better but i'm gonna try and see i'm gonna do some research and see how i can clean this up a little bit this is a different challenge i mean he can kind of scan rocks all day but when it comes to humans it is definitely a lot harder okay so if you have any questions let me know if you have any tips on how i can make this a little bit better also let me know and i'll see y'all later you

Channel: JSFILMZ

Views: 23,881

Keywords: 3d scan to ue5, reality capture tutorial for beginners 2021, how to 3d scan human in reality capture, unreal engine 5 photorealistic graphics, Unreal Engine 5 Realistic Graphics or Not, realistic 3d model ue5, photogrammetry unreal engine 5, unreal engine 5 3d scan, 3d scan unreal engine 5, photogrammetry ue5, reality capture to unreal engine 5, photogrammetry beginner tutorial 2021, Street photogrammetry with jsfilmz, reality capture, 3d Scan Human Head with Reality Capture

Id: rwUTuy9v4Wk

Length: 17min 42sec (1062 seconds)

Published: Wed Jul 14 2021