c172p
c172p copied to clipboard
c172p-team
Retopo the 3D model (or the 3D issues effort)
There are some 3D issues opened. Long time ago, i has noted a "retopo" on our old TODO file.
Well, the winter is coming. I want very much do that "retopo" right now, dealing ever it is possible, with the opened 3D issues related to the fuselage.
The bad thing about it, is that the model will finish with more vertices ( i don't know how much, but the vertcount will certainly increase).
The main goodness, is that adding more detail to the model will be easer, on the far future releases, when computer power and graphic card power, will increase anyway.
I was attracted before, by the idea of making small changes at a time. But now, i guess it will be very contra productive (at least, for me) because each change will fall in the process of "tweak all textures", "test", "correct", "tweak all textures", "pass".
Then, i guess that making the whole "retopo", these thing will be made in a blocked way. I prefer to correct some things in a row, rather than one thing and test, then other, then other, etc, etc, etc.
My mind will be switching less, if we proceed this way. I am focusing on the "retopo" now. Later i will be only testing and making corrections.
But, of course, the "retopo" will affect texturing, and i am not the only making textures.
@gilbertohasnofb, what do you think ?
There are two ways here.
(1) Ever i made a change, we need stop changing, tweak the textures, test, make possible corrections, and proceed.
(2) I will focus on the "retopo" only. After done, we both talk to choose the best way to cut and unwrap the mesh. Then we change to "texture issues effort".
I prefer the second way, Mainly because, after all, we will need to choose the best way to cut and unwrap the mesh, going through (1) or (2).
Obviously that the way (1) :
(a) Will duplicate tasks, because at the end we will need to decide how to cut and unwrap the mesh anyway. This will lead us to tweak the textures all again.
(b) Will delay the "retopo", because each change (or small blocks of changes) will stop the process to test, tweak, etc, etc.
(c) Will distract me a lot, because "retopo" and "unwrap" are separated processes and both consumes my brain (heheheheh !).
Waiting you opinion about.
Cheers to all.
If you want to retopo in order to make the model look better, then go for it! :smile:
Personally, I don't think a bit higher vertices count will (noticeably) reduce the fps, but we can always make an extra low-poly version (using the existing .ac model) before we release 3.8.
The bad thing about it, is that the model will finish with more vertices ( i don't know how much, but the vertcount will certainly increase).
A little bit more should be fine, but keep in mind that this is now the default aircraft in FG and we don't want to make it too heavy. There are plenty of folks out there who need a relatively light plane, so just don't go crazy in the details, particularly in small things like door handles, etc.
But now, i guess it will be very contra productive (at least, for me) because each change will fall in the process of "tweak all textures", "test", "correct", "tweak all textures", "pass". [...] But, of course, the "retopo" will affect texturing, and i am not the only making textures.
I agree, but are you sure we will need to do a lot of correction in the textures? As I said before, our paintkit is quite good, and so even if you break the fuselage into little parts you can still UV map it perfectly to our current paintkit. This way we may even get away without redoing the textures.
There are two ways here
Basically option 2 is better. Try to get the main retopo done and so then I can start tweaking all textures if necessary. But keep in mind what I wrote abote: a good UV map will avoid lots of work. Also note that we have some textures as xcf files, but the older ones are just a single layer in a png file, meaning that changing them is a pain in the neck. Please please please consider this when redoing the UV map.
I agree, but are you sure we will need to do a lot of correction in the textures?
The number of changes in the textures, is also proportional to the number of textures itself. If we have more textures, we need change more.
As I said before, our paintkit is quite good.
If you look better, it is not so good. There are a lot of issues related to the UV map.
When we changed the plane the first time, we tried don't change the livery system a lot. But we never talked about the layout itself. We didn't seat as a team and decided by the better way to unwrap the mesh. We decided by compatibility, not by the better layout.
The textures are really awesome. But we are in the same case when we started, because the front panel of the plane. The layout needed change. The difference is that in that case, this need was a lot obvious.
Take in mind that, changing shape will lead to change textures. Those processes are coupled.
Personally, I don't think a bit higher vertices count will (noticeably) reduce the fps...
Nor do i. But i planned to test it on the next weekend. Download flightgear and test the model with a plain white texture. IMHO, the increase of the texture size affects much more the performance that a little increase on the vertcount.
If you look better, it is not so good. There are a lot of issues related to the UV map.
Where? Which issues? The issues I see are much more related to not perfect UV mapping than to the textures themselves. If you look we have several opened issues about better mapping some parts of the plane using the same textures.
The textures are really awesome. But we are in the same case when we started, because the front panel of the plane.
But how is that going to change the paint kit?? Can't we just use the same ones we have? Surely you can UV map the new 3D objects using what we have, right?
IMHO, the increase of the texture size affects much more the performance that a little increase on the vertcount.
We offer both HD and non-HD textures, and you can change that during a flight. Offering two models on the other hand is a more delicate issue in my view. So yeah, I think we should keep the number of vertices as low as possible and not go crazy with little things. But that's just my 2 cents.
I think we must reuse the existing paint kit as much as possible, even if the model changes. @thevirtualfer, take into account redoing all textures is a hell of a work and they are currently very good! Try to uv-map to the same images and only change the map if it is really necessary. You can request help for this up-mapping!
In any case, I strongly believe we will need a new uv-map for the front of the aircraft. We are just extending artificially the image to cover the bow of the aircraft. Now, after a new detailed bow is modelled by @thevirtualfer, I believe this trick is not going to look very well any more. I guess a new front view must be added to the paint kit.
I guess a new front view must be added to the paint kit.
That's easy enough and I am willing to do that for all liveries.
@thevirtualfer also, if you are going to redo the fuselage around the nose, could you please have a look at this issue here: https://github.com/Juanvvc/c172p-detailed/issues/6
It is for me the absolutely worst problem we have with the 3D model, it makes the cockpit view look so bad and so different than in real life...
Where? Which issues? The issues I see are much more related to not perfect UV mapping than to the textures themselves.
Exactly what i am saying here. The UV mapping is not good. And doing a better unwrap will affect the textures.
Surely you can UV map the new 3D objects using what we have, right?
As much as possible. But no way to use it exactly as we have.
I think we must reuse the existing paint kit as much as possible, even if the model changes.
Again, it is exactly what i am saying. But being realist, you know that we will need change it. I am just preparing to the future. The process of unwrapping is not so near. But it will come and will be a headache to all of us. I am just trying to be sincere about this.
In any case, I strongly believe we will need a new uv-map for the front of the aircraft.
This is another example of how bad our unwrap is. That is all related.
take into account redoing all textures is a hell of a work and they are currently very good.
I am not saying the textures are bad or good. The paintkit is :
(1) The UV-mapping layout (2) The textures
The number (1) needs be rethinked. I know it may be frustating. As i said, we can try unwrap the mesh to almost fit into current layout. But honestly, i know that we will need change ALL the textures.
Gil.
I will help. Have you an account of the number of liveries ? Ignore the duplicated ones, because it is a matter of scale.
We can assign a half-to-half. I swear that i will do all the possible to not overcharge you.
Another way to deal with this is by making the same decision as before.
Create a new aircraft. A more detailed c172p.
But keep in mind what I wrote abote: a good UV map will avoid lots of work.
This is exactly what we don't have. It is all about i am saying. We need a better layout to avoid lots of work in the future. The layout must cover all sides of the plane (including bottom) and a special care is needed on junction points or curved surfaces.
Create a new aircraft. A more detailed c172p.
I am not sure I am up for that. Our model isn't bad by any means, and I'd much rather continue with it as it is. Also, if we have energy to create a new model then I propose that we do another plane (FlightGear is begging for a Cessna 182, a Piper Tomahawk or proper Cessna 152). But concerning this project here, given all the work we all invested I just think it would be crazy to redo the whole thing to be honest.
Just do the retopo and UV stuff in this project (in a feature branch). Near the end of 3.8 we can look at the performance and decide if we want to replace the .ac in Models/ or postpone it to 3.10.
It was only an option to put on the table.
But there are two points here.
(1) Our model isn't bad. But it is concerned to the personal opinion. Someone can say that our model is bad in the X or Y aspect.
(2) There will be not much difference on the amount o work using another model, or applying changes directly over this one. As before, we didn't make the plane from the ground. We got what was already made and added more things.
Basically, the work will be the same, doing the changes over this model, or using another model.
I give the suggestion, only because someone may be apprehensive by changing the default plane.
It is not my case.
I don't really understand the issue to be honest. @thevirtualfer just makes the model prettier by doing retopo + UV stuff + unwrapping + other cool Blender tricks :smile: Updates .blend and .ac files in a separate branch. After that we can decide if we just merge it or rename some of the files and have two -set.xml files (one pointing to the old .ac file and the other pointing to the newer/better one) Am I missing something?
You are right, @onox.
This is not the time to make this type of decision. I will proceed with the retopo. When finished, will talk about UV mapping.
Am I missing something?
HAHAHHAHA ! Yes, you are.
It is very very probable (in fact, it will occur) that the whole texture files will need changes. This will be a nightmare.
It is very very probable (in fact, it will occur) that the whole texture files will need changes. This will be a nightmare.
No, you told me that it's very probably that small modifications will happen, no that the "whole texture files will need changed". If a complete new paintkit will be created and the textures must be redone from scratch then I won't make any more textures for the plane, I can't just simply flush 6 months of work because you want to modify the 3D model. Because if I do that, then what will happen in 6 months from now? Will someone else come up with still a better "solution" still and recreate once again the 3D model and therefore flush once again my work? Sorry but that's a big no no to me, my time is really valuable to me.
Gil. You are upset. I understand this.
I said i will help with textures, didn't i ?
But ok. Lets look the cenario.
(1) Or we don't change the plane never more, because the paintkit can't be touched.
(2) Or we change the plane and tweak textures as needed.
If you want block plane modifications, it is ok to me.
Really, i understand your frustration. But plane mods will occur now or later.
Maybe the texture modifications will be not so hard to do. I don't know yet. Lets reach that point first. Then we talk about the UV mapping. I will not do anything without your "agreed".
If a complete new paintkit will be created and the textures must be redone from scratch
As i said, it isn't the case. I will try unwrap the mesh to best fit over what you have. But i know that it will not fit completely. We both can decide by the better cut.
No, you told me that it's very probably that small modifications will happen.
Not exactly what i said, but ok. The amount of modifications on the textures will propagate by the number of texture files we have. Maybe the problem is my english. But what i am saying is : If ONE texture file needs be modified (and definitely will need), then ALL THE OTHERS will need the same modification.
I can understand both the hesitation of @gilbertohasnofb to change (again!) the textures and the interest of @thevirtualfer in providing the best possible model. At the time being, I'm still confident that the improved model can be uv-mapped to the current textures quite well, except probably the front view.
@thevirtualfer, I'm going to the flying club next Saturday. Do you need any specific photo of the exterior of a c172? I think there are a couple of them in the hangars.
(I'm currently flying a Diamond DA40 and a Piper PA28 ;) )
I'm still confident that the improved model can be uv-mapped to the current textures quite well
God hear your words ! I am very pessimist about it. But of course, we will try to fit the maximum that we can.
Do you need any specific photo of the exterior of a c172?
Sure ? Great !
I am in trouble about the top and bottom view of the engine cowling. And most urgently, how the rudder connects with the vstab and the body. Googled a lot about this, without success.
In the bottom, it is basically the point the "NosewheelStruct" connects to the fuselage. There is a hole there ? What is the shape ?
I understand it all and we must see how it goes, but I made it clear that I don't think I will have time nor energy to recreate those same liveries from scratch due to a new paint kit. If we need to adapt the older ones, then of course I am in, but else it may be problematic situation. Anyway, let's see how these things go once the 3D model is fixed.
@thevirtualfer I do appreciate your concern with what I think and your offer of help. As I wrote, let's see how it goes once you have the model done.
Ok. I will proceed with the retopo. The fuselage increased the vertcount but doors, glasses, hstab, vstab and rudder, decreased. The steps are :
(1) retopo (doing good faceloops, using only quads, avoid poles, etc, etc) ;
(2) After done, clean the mesh ;
(3) After cleaned, is time for the problematic cut ;
I will make an account of the vertcount only after step (2).
But only to illustrate, look the door glass, for example.
Before retopo :
Can one really understand that ?
After the retopo :
A much more clean and easy to manage.
That definitely makes sense to me, being a total newbie to 3d stuff. While you retopo, would it be good to put these as reference background images in the .blend?
http://drawingdatabase.com/wp-content/uploads/2015/04/cessna_172_3v.jpg
The license is unclear, but this is very likely originates from the pilots operating handbook from Cessna.
I am sorry I am too new to blender at this point to help much more, but I tried placing the blueprint as background image for reference myself and initially the reference helps - to spot issues like the too low glareshield for example. However I did not manage to align it completely myself.
https://www.youtube.com/watch?v=-TQ-dBihGU0
I tried to do this but the model is a bit tilted compared to this one (tail low if I remember correctly), which is fine for side view since you can rotate the blueprint, but it distorts the top and front viewand you cannot really match the blueprint easily, thus I think I'd need to rotate the model a bit. Is it a lot of work to re-align the 3d model with the three-way? Having a good reference would be useful.
This is a completely new area to me, but it would be fun to learn, though poking at the 3d model likely requires a lot of coordination so that we don't mess each others work up :-)
Hi, @tigert !
I already tried with this blueprint, but our plane doesn't fit on.
Using a background image as reference is not the best approach (to me) because it shows only in the predefined views (top, left, camera, etc) and scaling and aligning it is a bit boring (to me).
I use "empties" to hold the blueprint. It gives more freedom in the alignment. Look :
To achieve this, first add an empty using "image" as the type.
Then, with the empty selected, go to here and set the image you want
Now, duplicate the empty and rotate as needed.
The distance of the empties are not important, if you use orthographic view.
You can put the empties on a separated layer that you can hide when they are disturbing what you are doing.
As i said, this blueprint is not in the same proportion of our cessna. Good blender to you, guy.
Guys. I am on this area of the plane, but as a rookie, i am in trouble with differences on reference photos.
What one of these is the shape of this part of the vertical stab for the 172p ? Or none ? Or any ?
Cheers to all, and thanks in advance.
Tuomas.
I started this effort trying to align the plane with blueprints. I have 7, but 4 are copies from the others. The remaining 3 different blueprints, don't match itself. For example, blueprint A doesn't match witch blueprints B and C, and so on. And none of the blueprints matches our plane (Sorry by my english).
But our plane almost match the fsx plane. The diffs are minimal.
I don't know how to proceed. But assumed that the proportions of our plane are correct.
If someone have a different opinion, let me know. The general format can be modified. It can be easy or difficult. I can't decide because i don't know the exact proportions. If someone with more confidence says :
- Use this reference as it is !
I will use. It will be no problem to me.
Cheers, bro.
@tigert.
I figured out why the blueprint doesn't fit. At least, i have a strong clue.
- The blueprint is so bad, that it contradicts itself.
We can check it, by drawing a red line that covers the tail in side view. See as it flows below the windows and finish in the spinner.
Now, the same red line in front view, shows that the blueprint is wrong with itself.
The red line is in the level with the windows and above the spinner.
The problem is that the author used some perspective rule that we will not guess.
Ideally, a blueprint needs to be orthographic. Without perspective involved.
Case closed, to this one.