The fact that there is the lower case "f" , AND there is already another word "Spy" has me absolutely bewildered, confused, and beside myself at your choice to pronounce it "Phspee".
Thanks for video. The only thing I want to say you needn't calculator in Blender to calculate such simple expression. If you want to input some ratio in some field you can print expression in the field (for example 3264/2448) and press Enter. The result 1.333333333... will be set in the field.
Can i just say I love how easily distracted you get, putting in the mirror for no reason just to teach us something. You are an amazing teacher and I appreciate the time you put into your videos. Thank you.
For anyone watching in the future.. the reason the scene is upside down when he imports it is because it’s upside down in fSpy. The Z axis is pointing down at 3:45. To fix it, change either your X or Y (depending which one you used to define left to right vs forward and back) to -x or -y in the drop downs in the top left corner.
This is a better tutorial for fspy than I've seen around. Fspy reads orientation from image metadata poorly (or very literally, depending whom you ask). If it's portrait and imports as landscape, even flipping it to horizontal in basic editor software won't rotate the image like you want when imported into fspy. Bring into GIMP (or similar), instead, and re-save it.
Happy to see 2.8 tutorials. 2.7x was so confusing to newcomers. This seems like the simplest way to do this, would love to see some more complex methods!
This is the type of person 2.8 dumped all the useful shortcut keys for. When you press E to extrude, you can just press X, Y, or Z without right clicking. You can also press shift + X, Y, or Z, and it will constrain to every axis other than the letter you press.
F - SPY (two words), the F is supposed to stand for Focal length. You 'spy' the focal length of the camera used to take an image. Fspy was made originally to exclusively just extract the focal length from a camera in order to then be able to feed that focal length to 3D camera tracking software where the focal length used was not known
Thank you very much CGMatter on your whole series about camera mapping and projections. You have the best tutorials about the matter: structure, content, examples, it's easy to follow :) And your pronounciation of "fSpy" really doesn't bother me at all ;) Just keep your pace and make some more excellent tutorials.
I see every time mind blowing tutorial on blender not like other modeling tutorials. I think this is what the case 😀 Please keep the tutorials going. Thank you.
Modifiers are non destructive, he has mentioned this in other videos. You should always try to use non destructive workflows, it allows you to easily fine tune your steps later.
The camera model and other stuff like settings could be in the EXIF data of the image (viewable in the image file properties) if take with a modern camera with the right options turned on. Couldn't hurt to check
@4:40 Your Z axis arrow (the blue one) is pointing down instead of up. That's why it imported upside down. To fix this you can click one of the "Vanishing point axes" on the sidebar and switch it to negative.
Very cool. Almost every other tutorial out there on camera mapping in Blender relies on the Project From View option for UVs. Being a long time LightWave user, I've missed the simplicity of its "Front Projection" mapping. This technique replicates that quite nicely. Thanks for the great tips!
Thank you for your awesome tutorials and channel! If you have time to answer, how would you go about adding objects to the scene by using a separate view layer for objects and 1 for a shadow catcher. I'm having problems on the shadow catcher side as I'm not sure if the answer is to raise slightly above the room floor plan?
Pretty sure you'd have all your objects (sphere + room + etc) on the first layer and only the shadow catcher on the second layer (at the height of the floor); the trick is then in all in compositing: you need to have 2 render layers node (one for the foreground and one for the background) and overlay them (using something like an alpha over node or mixRGB). Another option of course is to render out passes (diffuse, indirect diffuse, shadow, etc)
@@CGMatter Thank you for your help. I did attempt that but the shadow catcher in a separate view layer seems to dominate and cover the camera mapped floor creating an alpha shadow. I'm attempting it with the shadow pass now.
What a great tutorial, thanks for making it so easy. My imagination flew away and with it a question, if after raising the mapping as you explained it well, you want to project each of the walls in an interior 360 installation, how would you reder each face of the cube?
Great tutorial, I just want to say that blender can calculate things - just put 3264/2448 straight into the aspect box and it will give you 1.333. Also, you will almost never find a photo that's taken in a weird aspect ratio like 1.333...:1 - this is equivelant to 4:3
I already knew about this mechanic. But I see that you have a way better way of doing it. Before I knew about that modifier I would just go to camera view and "UV MAP FROM VIEW". It works the same way but requires you to be within camera view. But your method is more effecient as it corrects itself when you add more modifiers like subduvision. I also tend to make the geometry "shadow catcher" for CGI shots so I stil have the real footage behind the model but if the model is reflective it will stil reflect the mapped model with the textures.
How would you then display shadows if you were to add lets say a cube and a lightsource into the room. I did that and there are no shadows appearing on the floor/walls. Great tutorial btw!
ps: just found it out 1 minute after writing the comment lol! go to your texture of the room. Place a principal bsdf node inbetween the image texture and the material output.
I was wondering is ther anyway to use Fspy with Archipak or Arcimesh...so that exact dimensions of interior can be maintained. Problem i'm having is when I bring in fspy camera and start modeling, the height of the walls extruded on the x axis is enormously too large. Average height of real wall is 2.7 meters but ...I can seem to get fspy to maintain those measurements...Please help if you now of any workaround
Great Tutorial..i just tried it and got satisfied with the results...but i have a question that does Camera projected mesh affected by any artificial lights or not..i am trying to get a bit more edetail on the projected mesh but ponit light,area light,sun does not affected on the mesh..
This is cool, but what I really want to do is take the render from one camera in the scene and use it as a texture map for one of my materials. (So I don't have to render it out to an image and then import that image into a texture, allowing me to make tweaks to the image on the fly or animate the textures within the file itself.)
I download the add-on for blender and this one not work in blender 1.80 (v2.80.75). I can't find it in add-ons dialogue window . Any advice? Thank you.
Although I admire CGMatter videos, a couple of mistakes were made in this video. 1/ First of all this image has a 3-point perspective and you could have leveraged this by setting "Principal Point" to "From 3rd vanishing point". fSpy would determine the principal point as well as focal length from the image. This is why your Z axis doesn't match well in Blender. 2/ For 2 vanishing points you should adjust the principal point manually to align the Z axis 3/ In case of 2 vanishing points you don't set the focal length in the bottom right corner, but rather by moving the principal point left or right (it's a difficult process). 4/ You could have placed the other X line at the ceiling level for better accuracy 5/ Although not crucial in this case, it would help to set the origin on the floor, set the reference distance along Z, and then start modeling in Blender from the origin. I talk about these (and all other fSpy options) in my tutorial: th-cam.com/video/daiMOYR8GS8/w-d-xo.html and present how to deal with non-trivial images having various types of perspective.
Doing a project from view (looking from the camera), works better in my opinion, you don't need that many faces and gives you more control over where the stuff goes
Hello, this tutorial is making me excited but I'm not fully understanding it. What I'm trying to do is make simple building facades from already perspective-corrected textures of building facades. There are many tutorials on this already, and it's easy to do with loop cuts and extrudes for simple square shapes. My problem is that some of my facades have curves and angles on them and when I start to play with that geometry, the texture distorts. I want to be able to move my geometry around, divide, extrude, bevel without the texture really changing, basically as if there was a projector projecting the texture from a fixed point and then I can wiggle the geometry to perfectly match the unchanging projected texture. Is this possible? Like in your example your whiteboard has a little tray along the bottom. How would you add that tray? Could you add some edges, extrude out the tray and wiggle it around until it lined up with the texture rather than the texture moving/warping the moment you moved a face?
Here is a tutorial I made on it: th-cam.com/video/8NYNiayHvJI/w-d-xo.html generally baking is fairly straight forward but using projected textures requires some extra work (covered in the linked tutorial). Hope it helps :)
Probably it doesnt work in 2.8, but for 2.79 was addon called BLAM that did the same thing without need to use external software, marking the lines with grease pencil.
Possibly, however this technique generalizes to any type of geometry. The one by one approach will become super tedious when you need to handle the projection of 200+ faces.
How would you go ahead and map a scene with multiple pictures taken? Let's say, I'd like to do something like photogrammetry but manually by using camera projections to get cleaner results with less poly count. Is something like that possible?
Some things that come to mind is creating a roughness map so things like the white boards are shinier than everything else, you can also generate a bump map from the original photo but it should be made very subtle since it's not physically correct at all. Best tip I have is to add very subtle mist (fog) to the scene which gives it tons of volume.
Hi CGM, I am not able to find a tutorial on how to 2d/3d animate over a video (as a background) in real-time in blender 2.8 , can you make a tutorial for the same?
Hi there, thanks for this video. When I worked my way through it with my own files, your mirror effect worked perfectly. However, when I tried to add an object into the scene to cast shadows onto my 'walls', the shadows wouldn't not appear. Is that just a basic conceptual misunderstanding on my behalf? That projected textures can't receive shadows? I've tried to look for information on this, but don't have the vocabulary to ask the right questions!
In the tutorial there is no proper shader on the material. If you add a shader and plug the image into the diffuse slot, the material will be able to receive shadows. Speaking of which, the room will now have shadows from the walls, and you will probably need some light inside.
@@tiagotiagot I'm not sure, unfortunately. I only know this because BLAM's GitHub page has a notification saying that it's deprecated, with a link to fSpy as the alternative. They're made by the same developer I believe.
The fact that there is the lower case "f" , AND there is already another word "Spy" has me absolutely bewildered, confused, and beside myself at your choice to pronounce it "Phspee".
I keep laughing when he says it. Like how in the absolute fuck did he go with that pronunciation lmao
Yeah I like Eff-Spy more but I can't fault anyone for just guessing and moving on, no big whoop.
@@swashbucklingmonkey I mean F spy literally means to basically "spy"/find out the f (focal) point lol
@@rolithesecond That's not entirely obvious to everyone out there.
plot twist: it's really F's pee xD
Thanks for video. The only thing I want to say you needn't calculator in Blender to calculate such simple expression. If you want to input some ratio in some field you can print expression in the field (for example 3264/2448) and press Enter. The result 1.333333333... will be set in the field.
just don it in your head
@@DestianLight My head is not so smart. :-D
@@Avgur_SmileJust practice a little...
@@DestianLight once you get good at reducing fractions it's just a matter of memorisation :D
@@toadbard Sure and when thet technology is broken or gone or war or something. Back to stone age I suppose...
Can i just say I love how easily distracted you get, putting in the mirror for no reason just to teach us something. You are an amazing teacher and I appreciate the time you put into your videos. Thank you.
Nice. Be aware that the "Z" trick works with extrude too. You don't have to e,esc,g,z you can just e,z and move the mouse.
that sounds ez
For anyone watching in the future.. the reason the scene is upside down when he imports it is because it’s upside down in fSpy. The Z axis is pointing down at 3:45. To fix it, change either your X or Y (depending which one you used to define left to right vs forward and back) to -x or -y in the drop downs in the top left corner.
Thanks
@@nicholasnewman4572 No problem, happy blending
Thank you!
This is a better tutorial for fspy than I've seen around. Fspy reads orientation from image metadata poorly (or very literally, depending whom you ask). If it's portrait and imports as landscape, even flipping it to horizontal in basic editor software won't rotate the image like you want when imported into fspy. Bring into GIMP (or similar), instead, and re-save it.
Happy to see 2.8 tutorials. 2.7x was so confusing to newcomers. This seems like the simplest way to do this, would love to see some more complex methods!
This is the type of person 2.8 dumped all the useful shortcut keys for.
When you press E to extrude, you can just press X, Y, or Z without right clicking.
You can also press shift + X, Y, or Z, and it will constrain to every axis other than the letter you press.
F - SPY (two words), the F is supposed to stand for Focal length. You 'spy' the focal length of the camera used to take an image. Fspy was made originally to exclusively just extract the focal length from a camera in order to then be able to feed that focal length to 3D camera tracking software where the focal length used was not known
i dont think anyone in human history has ever said fspee
@@commodore7331 except this guy apparently haha
Thank you very much CGMatter on your whole series about camera mapping and projections. You have the best tutorials about the matter: structure, content, examples, it's easy to follow :) And your pronounciation of "fSpy" really doesn't bother me at all ;) Just keep your pace and make some more excellent tutorials.
I see every time mind blowing tutorial on blender not like other modeling tutorials. I think this is what the case 😀
Please keep the tutorials going. Thank you.
also could use editMode->u->projectFromView, but the uv project modifier is cool too, didnt knew it existed 👍
It's a lot easier in my opinion and less geometry
@@kendarr I agree, he made this unnecessarily long so that he can get more money.
UV project is way better and easier
@@unicornhuntercg I don't think hes a gredy guy, he probably didn't knew the easier way
@@unicornhuntercg no bro. The video was already 9 minutes long
Modifiers are non destructive, he has mentioned this in other videos. You should always try to use non destructive workflows, it allows you to easily fine tune your steps later.
This is the first video of you I have seen. Now I've seen every single video you have made so far, on both of your channels
Love this style of tutorial before the descent into madness...
XD
fspy is properly pronounced "jeff".
🤣
@@CGMatter lmao
blessed comment
it's actually geff
just stumbled upon your channel and i loved the clarity of the vid and the knowledge, was looking for this kind video actually. SUBBED!
Really appreciate it! Thank you for watching ;)
The camera model and other stuff like settings could be in the EXIF data of the image (viewable in the image file properties) if take with a modern camera with the right options turned on. Couldn't hurt to check
you can also use the “n menu” > view > camera lock > lock camera to view, in order to move your camera easily
@4:40 Your Z axis arrow (the blue one) is pointing down instead of up. That's why it imported upside down. To fix this you can click one of the "Vanishing point axes" on the sidebar and switch it to negative.
Maybe you want to pick the area with the highest contrast to align your axis...
Such a cool technique. Thanks for the great tut!
Glad you liked it 😊
9:51 - you can get image size from image node porps - select node image and hit "n" key
Very cool. Almost every other tutorial out there on camera mapping in Blender relies on the Project From View option for UVs. Being a long time LightWave user, I've missed the simplicity of its "Front Projection" mapping. This technique replicates that quite nicely. Thanks for the great tips!
Thank you for your awesome tutorials and channel! If you have time to answer, how would you go about adding objects to the scene by using a separate view layer for objects and 1 for a shadow catcher. I'm having problems on the shadow catcher side as I'm not sure if the answer is to raise slightly above the room floor plan?
Pretty sure you'd have all your objects (sphere + room + etc) on the first layer and only the shadow catcher on the second layer (at the height of the floor); the trick is then in all in compositing: you need to have 2 render layers node (one for the foreground and one for the background) and overlay them (using something like an alpha over node or mixRGB). Another option of course is to render out passes (diffuse, indirect diffuse, shadow, etc)
@@CGMatter Thank you for your help. I did attempt that but the shadow catcher in a separate view layer seems to dominate and cover the camera mapped floor creating an alpha shadow. I'm attempting it with the shadow pass now.
Thank you so much for sharing this video. This toturial is so helpful 🤗🤗👌👌
This is really awesome :D
thanks for the quick tutorial :)
What a great tutorial, thanks for making it so easy. My imagination flew away and with it a question, if after raising the mapping as you explained it well, you want to project each of the walls in an interior 360 installation, how would you reder each face of the cube?
Fspy was once an addon for blender called Blam :), it works much better now thats off blender
Your videos are sooo amazing I wish I found them earlier
Amazing tutorial man :) subbed for more like this
Thank you! Got a cool one coming up, stay tooned :)
4:18 why you don't pick contract between white and black sides and use hard way instead :(((
Its F spy, not efspee
then why is it called "fspy" and not "f spy" mah boi cgmatter is correct on this one
source i am an f spy
@@blackwhattack Go back to your eephone then 😝
Great tutorial 😁
Thanks for watching :)
You used the wrong mode, you only had 1 vanishing point but used the option for 2. Still worked but it might have been easier in the right mode.
Also a weird choice of points, I mean there is a whiteboard right there
Great tutorial, I just want to say that blender can calculate things - just put 3264/2448 straight into the aspect box and it will give you 1.333. Also, you will almost never find a photo that's taken in a weird aspect ratio like 1.333...:1 - this is equivelant to 4:3
Thank you , I want to know Camera Projection can transfer to UE5 ?
Very nice, thank you
I already knew about this mechanic. But I see that you have a way better way of doing it. Before I knew about that modifier I would just go to camera view and "UV MAP FROM VIEW". It works the same way but requires you to be within camera view. But your method is more effecient as it corrects itself when you add more modifiers like subduvision.
I also tend to make the geometry "shadow catcher" for CGI shots so I stil have the real footage behind the model but if the model is reflective it will stil reflect the mapped model with the textures.
phspee! :)
This is brilliant!!!
Great tut! Thank you for sharing!
When I look up fspy in the addons its not there, maybe im searching the wrong name?
same
How do make the material sensitive to light and shadows?
Shizzle! Bro this if hot. Thank you.
How would you then display shadows if you were to add lets say a cube and a lightsource into the room. I did that and there are no shadows appearing on the floor/walls. Great tutorial btw!
ps: just found it out 1 minute after writing the comment lol! go to your texture of the room. Place a principal bsdf node inbetween the image texture and the material output.
I just call it "ef spi"(written as pronounced)
eff spee?
@@longpinkytoes eff spai
Huh, I pronounce it as /ɛf spaɪ
/
written as pronounced doesn't apply in english
I pronounce it "eff SPAH SAPPIN' MAH SENTRY!".
I cant open fspy. What program do I need to open a PY-file? there isnt a application in the zip too, so how do I even open the program?
I like this video. I subscribed.
can we project image sequence of a moving cam projected?
Very cool, thank you!!
I was wondering is ther anyway to use Fspy with Archipak or Arcimesh...so that exact dimensions of interior can be maintained. Problem i'm having is when I bring in fspy camera
and start modeling, the height of the walls extruded on the x axis is enormously too large. Average height of real wall is 2.7 meters but ...I can seem to get fspy to maintain those measurements...Please help if you now of any workaround
Very nice tut, but you mapped X axis wrong on fSpy, make sure to keep Z pointing up, and you wont need to rotate the objects later.
Great Tutorial..i just tried it and got satisfied with the results...but i have a question that does Camera projected mesh affected by any artificial lights or not..i am trying to get a bit more edetail on the projected mesh but ponit light,area light,sun does not affected on the mesh..
When I opened prefrences and searched it up. It didn’t show the fspy addon, even I started it again
This is cool, but what I really want to do is take the render from one camera in the scene and use it as a texture map for one of my materials. (So I don't have to render it out to an image and then import that image into a texture, allowing me to make tweaks to the image on the fly or animate the textures within the file itself.)
same here i subbed too...gr8 more blender tuts
I download the add-on for blender and this one not work in blender 1.80 (v2.80.75). I can't find it in add-ons dialogue window . Any advice? Thank you.
nice tutorial, when I install the fspy add-on it still doesnt show up in the import options. Is it not working in the most recent 2.8 release?
Although I admire CGMatter videos, a couple of mistakes were made in this video.
1/ First of all this image has a 3-point perspective and you could have leveraged this by setting "Principal Point" to "From 3rd vanishing point". fSpy would determine the principal point as well as focal length from the image. This is why your Z axis doesn't match well in Blender.
2/ For 2 vanishing points you should adjust the principal point manually to align the Z axis
3/ In case of 2 vanishing points you don't set the focal length in the bottom right corner, but rather by moving the principal point left or right (it's a difficult process).
4/ You could have placed the other X line at the ceiling level for better accuracy
5/ Although not crucial in this case, it would help to set the origin on the floor, set the reference distance along Z, and then start modeling in Blender from the origin.
I talk about these (and all other fSpy options) in my tutorial: th-cam.com/video/daiMOYR8GS8/w-d-xo.html and present how to deal with non-trivial images having various types of perspective.
its so much better than doing this in non 3d softwares lol no offence
Doing a project from view (looking from the camera), works better in my opinion, you don't need that many faces and gives you more control over where the stuff goes
Hello, this tutorial is making me excited but I'm not fully understanding it. What I'm trying to do is make simple building facades from already perspective-corrected textures of building facades. There are many tutorials on this already, and it's easy to do with loop cuts and extrudes for simple square shapes. My problem is that some of my facades have curves and angles on them and when I start to play with that geometry, the texture distorts. I want to be able to move my geometry around, divide, extrude, bevel without the texture really changing, basically as if there was a projector projecting the texture from a fixed point and then I can wiggle the geometry to perfectly match the unchanging projected texture. Is this possible?
Like in your example your whiteboard has a little tray along the bottom. How would you add that tray? Could you add some edges, extrude out the tray and wiggle it around until it lined up with the texture rather than the texture moving/warping the moment you moved a face?
thanks
Thanks for the vid!
Please upload the Files used in the Video so that we can follow everything --- from absolute beginner
whats the program that puts what you write on the top left
I think its called f - spy, not fspy as a one word.
f stands for focal length, so its a focal length spy ;)
what about frisbee? :P
@@longpinkytoes what about it?
@@suspiciousstuff4999
good pronunciation for fspy
@@longpinkytoes 😭😭😭
I hope he's just trolling us.
how to bake textures? so that I can import a mapped object in a software like after effects through element 3d plugin
Here is a tutorial I made on it: th-cam.com/video/8NYNiayHvJI/w-d-xo.html
generally baking is fairly straight forward but using projected textures requires some extra work (covered in the linked tutorial). Hope it helps :)
This is soo awesome, imagine what game devs can do with that!
Sinan Akkoyun Game devs are way ahead bro.
Legend 🙌
ok at 0:00 , click and hold your left arrow button
and MEDIA OFF LINE
Probably it doesnt work in 2.8, but for 2.79 was addon called BLAM that did the same thing without need to use external software, marking the lines with grease pencil.
blenderartists.org/t/retiring-blam-and-announcing-fspy-a-stand-alone-camera-matching-app/1132588
Vanishing point error?, could it be cuz it was set to 2 point vanishing?
Using the uv modifier seems way harder than just uv unwapping from view..
The first frame of the video is "Media Offline" lol
I have a macbook air, does Fspy work with this?
Wouldn't it be better to take the individual planes and "project from view" the floor, left wall, right wall, and ceiling?
Possibly, however this technique generalizes to any type of geometry. The one by one approach will become super tedious when you need to handle the projection of 200+ faces.
what if your aspect is lower than 1
How would you go ahead and map a scene with multiple pictures taken? Let's say, I'd like to do something like photogrammetry but manually by using camera projections to get cleaner results with less poly count. Is something like that possible?
2:18 the sex axis?
LMAO
anyone see the "media offline" text at 0:00?
where did find that reference photos?
Some of the best blender tutorials out there on Your channel. Just get a better microphone to perhaps get the voice away from that knife sharp sound?
Just got a new microphone, will try it out for the next video :)
Thank you
Thanks for watching :)
I have a question: how to export to fbx, obj with the photo mapping (as texture)? If not, it seems there is no value to create a work like this...
Every time I load an image into fspy it rotates it. How do I fix this??
i miss the old cgmatter D:
Can you give volume to some details using the same picture, without modelling them? Some texturing/shading technique?
Some things that come to mind is creating a roughness map so things like the white boards are shinier than everything else, you can also generate a bump map from the original photo but it should be made very subtle since it's not physically correct at all. Best tip I have is to add very subtle mist (fog) to the scene which gives it tons of volume.
@@CGMatter Thank you very much, I'll do some reading about those ideas.
please make a video about the harder case where there's objects like a table or couch in the photo. IDk how to do that yet :/
how can i bake the projected texture into a normal uv space(another unwrapped channel) ?
what about a forest scean?
how do I take a Photo of what jeg have make ???
when you extrude (hitting E), just hit X, Y or Z instead of canceling
Hi CGM, I am not able to find a tutorial on how to 2d/3d animate over a video (as a background) in real-time in blender 2.8 , can you make a tutorial for the same?
Camera mapping onto a sphere?
Hi there, thanks for this video. When I worked my way through it with my own files, your mirror effect worked perfectly. However, when I tried to add an object into the scene to cast shadows onto my 'walls', the shadows wouldn't not appear. Is that just a basic conceptual misunderstanding on my behalf? That projected textures can't receive shadows? I've tried to look for information on this, but don't have the vocabulary to ask the right questions!
In the tutorial there is no proper shader on the material. If you add a shader and plug the image into the diffuse slot, the material will be able to receive shadows. Speaking of which, the room will now have shadows from the walls, and you will probably need some light inside.
cant find the addon
Dude this is making me high. Good dope.
Why not use the BLAM addon? It's not available for 2.8 yet?
fSpy is the successor to BLAM, actually!
@@editfoxextras8530 Really? Why did it move out of Blender?
@@tiagotiagot I'm not sure, unfortunately. I only know this because BLAM's GitHub page has a notification saying that it's deprecated, with a link to fSpy as the alternative. They're made by the same developer I believe.