Forum

User posts Marcos Calvi
15 December 2016 16:18
Hi there:

I have reinstalled Windows 7 after a hard disk failure and I had to do it to a new HD, I could not export a backup of the projects I was working in. After reinstall I was able to access the blend4web folder of the old installation and copy the Blender, apps_dev and assets folders to the new install of b4w. When I open the files in Blender it works ok and even the server is serving the apps but the problem is no project is listed in the project manager page. Do you know how can this be? It was me putting the old files that broke the project manager? I thought it was enough to drop the folders in the installation. Do I have to check other file or something to get the projects working in the project manager?

Any help is welcome…
24 October 2016 21:29
You can download the file at this link
If you press 9 or H you get the keys assignment.
Hope it works…
24 October 2016 21:17
It is 16.09.2. I've been trying a little more and it is strange, if I apply the rotation to the other axis it works ok and as I roll the ship the camera rotates up and down or left and right depending on the axis I apply the rotation to but not to the Z axis…
21 October 2016 20:45
Hi:
But if I understand you right, you want roll rotation (wiki), which is rotation around z axis in camera's local coordinate space.

Yes, it is roll rotation. Along the local z axis of the camera. What I've found is that when I use the y parent rotation and add it to the quaternion via rotateAxis if I use any of the other axis works and the camera rotate but in the wrong direction.

I keep trying…
21 October 2016 02:05
Hello again:

I've been trying to rotate a STATIC type camera along the y-axis, the front axis, with the new way to call it. (up-down, x-axis, left-right, z-axis). Well, I wanted to rotate around and I have found is imposible, at least to me.
I have tried via a quaternion and the transform api:
                var _cam_quat = m_trans.get_rotation(_cam, _vec4_tmp2);
		m_quat.rotateY(_cam_quat, _my_rot_y, _cam_quat);
		m_trans.set_rotation_v(_cam, _cam_quat);

If I use the other axis it works ok, but with the y axis doesn't work. Is this a limitation of the camera or I am doing something wrong?
15 October 2016 17:03
Hi again:

Yes, the character thing worked, thank you I was seeing me changing all the character movement to rigid body…Thanks Evgeny.

Well, the thing seems to be in his way for now, I have down-scaled ten times and port it to 16.09, now I am going to correct the movement speed so it must be ten times slower now, then I'll upload a file so if someone wants to take a look can.
14 October 2016 00:17
Hi again:

This is the link to the file. I hope you can download it without any problem.

If you press 9 or H you can see the keys. The ship is a rigid body so if you hit anything you can start spinning without control. The roll over the z-axis is not followed by the camera (one of my problems). Ah, and it is still working in 16.08 so the axis is the old way. In a couple of days I will go updated and will make a new version.
13 October 2016 02:06
Hello again:

I've been working in this a little more and the behaviour is intriguing. The thing is when you are reaching more or less 5000 meters the parenting of the objects starts to mess up. It is interesting that if you return to the origin the things get ok and no strange behaviour happens.

Now I think the solution is pretty obvious although I have not tried it: working with smaller objects . I'm going to try to work with objects 10x smaller so the impression of the space would be 50 kilometers, which is pretty big if you adjust the speed of the ship and fake the units.

Tomorrow I'm going to upload the file, as I want to include the keys assignment in a little help div so you can look at the problem for yourself.
12 October 2016 15:04
Hi there:

I've been working in a project which implies really big scenes, I'm trying to make kind of a space voyage game and I'm using the Gliese Stars Database to create the stars programatically in blender with Python which is great and in a couple of minutes you have the 1500+ stars in the scene. If you do it treating the coordinates units as meters it works fine and all the stars appear correctly in the exported scene and you can fly by and everything, but…

If you increase the distance by multiplying the distance you find one pretty bad problem I didn't have notice of. Indeed I'm not quite sure if the problem is what I think it is. I think it is about the precision of the calculations when you deal with big numbers. The thing is: as the 'space ship' is a cabin wich I attach the camera to by a semi stiff constraint of the camera to an empty that is at the same time a child of the rigid body which is the hull of the ship, when you progresively go further in space the constraint starts to behave in strange ways, in fact it ends up losing completely the position and no longer works properly, even going out of the cabin and viewing the ship from outside.

Then I found this page in the Blender documentation which explains that as we are using float 32 numbers the calculation with big numbers loses precision as we get further away from the origin, a thing a have heard of but never suffered as I hadn't work with so big distances.

Googling a little I found that this is not only a Blender problem but a problem with all the game engines and 3d software and hardware that are using these mumbers in it's core (which is pretty astonishing, I hope my bank to use double precision, hehe…)

Aside from that I'm still checking if this is in fact the case cause I'm using blender parenting and blend4web parenting at the same time and it could be that using only one type it works ok.

I tried to scale the scene too. In the blender documentation says that if you don't want so much precision in sculpting et al and scale the blender units to be 10 meters instead of 1 meter you get better calculations at big distances. I tried it, and it seems to work better, you go further away and the constraints keep stable.

But I found another problem the character don't seem to get the rescaling well and although the object itself gets scaled the physic body is still the same size and the body floats one meter above the ground

Is there any workaround to this problem? I'm still working on it and maybe I can adapt the object ok in the javascript using b4w api. It appears to be a hard problem. I mention it here because it is an issue that is pretty interesting all that float 32 thing…

Well, if you have any idea that could work it will be welcome, I'll keep on it and update this post as I go through it.

Cheers.
28 September 2016 13:24
Aaaah, all right, I'll try those hints. I thought it had something to do with the environment of the variables, so I was lost, this opens other way, thank you.