Show Posts
1
Scripting and Reverse Engineering / [FF8] Camera Research / reconstruction
« on: 2019-06-26 11:59:44 »
Hello guys!
It's been a while since the last time but the "remaster" announcement woke me up from a long sleep
What's the point of this topic and why should it be so important?
Cameras are one of the not yet fully documented things in FF8 but they could provide great help to the 3D reconstruction of many stuff that could lead to new/better 2D background mods for example.
Also later on, when the remaster will be released (if the engine is going to be moved to Unity or something), maybe an actual 3D environment could actually be injected in the game (or replaced for some battle stages for example).
We already can use walk-meshes for having some spatial reference and they helped me a lot to understand the actual depth for some areas.
I'm sure those mesh come from the actual 3D model since some of them kept their relative position (you can notice that if you import most of the garden ones..the central hall ones just overlap each other with no need to move them)
Having exact cameras views in the 3D would help with precise reconstruction and be used for texture camera projection.
I guess we could try to make some sort of inverse-photogrammery to get some raw-3d models of some stuff.
This is what I'd like to get in a more precise way (I adjusted the camera position manually to get this)
Even better would be having a precise reconstruction of the camera movement during cinematics with 3d characters overlay (we need to figure out how to read those data)
So far this is what I think I understood:
-walk-mesh come from the original 3D models then they're a reliable reference of the 3D environment
-the in-game 3D camera probably "render" a bigger region which is later cut from the game (the camera doesn't actually move when you see the background slide..it just show a different slice of the same camera render giving the illusion of movement while keeping the character aligned on the background) ---> see camera ranges
So far this is what I'm trying to understand and I'm quite sure about:
Camera position read from Deling doesn't actually match the walkmesh but if you swap the order the position you get is probably correct (X->Y , Y->Z , Z-> X) ..otherwise all the walkmesh position are the one to be swapped.
Camera orientation from my guess is guided from 3 vectors (even though probably just 2 of them should be enough to get the orientation)
1st Vector represents the axis pointing to the "right" of the camera
2nd Vector represents the axis pointing to the "bottom/down" of the camera
3rd Vector represents the axis pointing to the "front/aim" of the camera
So far this is what I'm trying to understand but I really don't know how to handle:
-In the very first "in-game cinematic" there must be some sort of object masking away the characters when they are "behind" the column... I would guess that should be some sort of mesh (similar to the one for the walk ) that serves the purpose of hiding 3d models... could be that or some sort of 2d mask tracked but I hope for a 3D plane that could give more spatial info.
see here (5:57) : https://youtu.be/0Gfcjm4a1Lw?t=349
-I don't get how the "zoom" value works and how I'm supposed transfer it to a 3D software (like 3dMax or Maya)
-We don't know much yet about camera animation data but the number of "positions" (??) doesn't match the supposed number of frames in the cg animation.. we need to figure out how those data are supposed to be read..!
It's been a while since the last time but the "remaster" announcement woke me up from a long sleep
What's the point of this topic and why should it be so important?
Cameras are one of the not yet fully documented things in FF8 but they could provide great help to the 3D reconstruction of many stuff that could lead to new/better 2D background mods for example.
Also later on, when the remaster will be released (if the engine is going to be moved to Unity or something), maybe an actual 3D environment could actually be injected in the game (or replaced for some battle stages for example).
We already can use walk-meshes for having some spatial reference and they helped me a lot to understand the actual depth for some areas.
I'm sure those mesh come from the actual 3D model since some of them kept their relative position (you can notice that if you import most of the garden ones..the central hall ones just overlap each other with no need to move them)
Having exact cameras views in the 3D would help with precise reconstruction and be used for texture camera projection.
I guess we could try to make some sort of inverse-photogrammery to get some raw-3d models of some stuff.
This is what I'd like to get in a more precise way (I adjusted the camera position manually to get this)
Even better would be having a precise reconstruction of the camera movement during cinematics with 3d characters overlay (we need to figure out how to read those data)
So far this is what I think I understood:
-walk-mesh come from the original 3D models then they're a reliable reference of the 3D environment
-the in-game 3D camera probably "render" a bigger region which is later cut from the game (the camera doesn't actually move when you see the background slide..it just show a different slice of the same camera render giving the illusion of movement while keeping the character aligned on the background) ---> see camera ranges
So far this is what I'm trying to understand and I'm quite sure about:
Camera position read from Deling doesn't actually match the walkmesh but if you swap the order the position you get is probably correct (X->Y , Y->Z , Z-> X) ..otherwise all the walkmesh position are the one to be swapped.
Camera orientation from my guess is guided from 3 vectors (even though probably just 2 of them should be enough to get the orientation)
1st Vector represents the axis pointing to the "right" of the camera
2nd Vector represents the axis pointing to the "bottom/down" of the camera
3rd Vector represents the axis pointing to the "front/aim" of the camera
So far this is what I'm trying to understand but I really don't know how to handle:
-In the very first "in-game cinematic" there must be some sort of object masking away the characters when they are "behind" the column... I would guess that should be some sort of mesh (similar to the one for the walk ) that serves the purpose of hiding 3d models... could be that or some sort of 2d mask tracked but I hope for a 3D plane that could give more spatial info.
see here (5:57) : https://youtu.be/0Gfcjm4a1Lw?t=349
-I don't get how the "zoom" value works and how I'm supposed transfer it to a 3D software (like 3dMax or Maya)
-We don't know much yet about camera animation data but the number of "positions" (??) doesn't match the supposed number of frames in the cg animation.. we need to figure out how those data are supposed to be read..!
texture 0
texture 1
texture 2
texture 3
Text0
Text1
