Ordnance Updates

[Zorin]

1. Model orientation in Max was wrong. The model should be placed in max so it's nose points downwards on the Max Top view. The CG of the object should be placed at 0,0,0 coordinate.

How am I supposed to know that?

2. Texture size was too large (1024x1024) while the texture itself was not quite detailed. No point in using large texture, if one doesn't plan to paint anything on it.

I always paint and setup textures in a higher resolution. It was just the base setup and had no detailing yet. Of course it would be resized for the finished product...

3. Texture mapping wasn't quite good and savvy. Too many stitches in UW makes painting details on texture difficult.

I UVW to reduce possible distortion even if that requires a little more time for proper texture painting. Besides, anything that would have had seams you removed anyway...

4. Thin parts, like fins, should be made, using 2-sided material, their shape can be easily "cut" from a simple rectangle, using alpha channel.

Why should I do it if I can keep higher detail within the tri-level limit I have been given? I also was told that I would not have to build for 2-sided material, but could use flipped faces and only count one side per part. Could you guys please make your mind up....

Alpha channels are just to be avoided as far as I am concerned as they only allow for very crude results on this resolution levels.

5. No LODs found in the file.

Surprise, see point one. If I don't get proper info on that, how can I include it? Besides this was clearly labelled as a LOD0 preview...

6. No collision object found in the file. Ordnance doesn't need mesh collision, it is enough to use collision sphere which is just default Max primitive, 32 sides sphere, whose diameter is slightly larger, then ordnance diameter without fins. The CSphere object should be placed at the 0,0,0 coordinate.

See point 1 and 5.

7. TGA file should be used as texture.

Noted.

I did a brief remodel of one of the torpedo variants ... without any loss to visual quality

Cute euphemism and the detonator is not placed where you did "correct" my detonator to. Believe it or not, I actually know what I am doing. Also, the variants are meant to allow every type of torpedo plane to use the proper torpedo tail for any given time frame. If you don't want the four, which could easily be 8 btw, then just leave out the tail assemblies altogether, cause anything else would be a halfarsed job.

But I am glad that the community now has a comparison. I will gladly dumb down the meshes, anyway. Anything is better than the standard crap that is in game at the moment.

[EJGr.Ost_Caspar]

Hm... just came to my mind... do you need example models? Or do you have already?

Pls don't take poly limits as something to exhaust at any costs - 500 is a number for comparible complex objects. Every saved polygon is a good polygon.

Sorry if we look unsatisfied with the status (LoDs etc.) ... we took it, that the model was finished - at least the one, who wrote the report.

[Zorin]

Originally Posted by EJGr.Ost_Caspar

Hm... just came to my mind... do you need example models? Or do you have already?

Pls don't take poly limits as something to exhaust at any costs - 500 is a number for comparible complex objects. Every saved polygon is a good polygon.

Sorry if we look unsatisfied with the status (LoDs etc.) ... we took it, that the model was finished - at least the one, who wrote the report.

Examples sure wouldn't hurt, cause those would give me the file structure you want and an insight on how you want to see things done cause it is apparent that that is the only way you accept them.

Well, I usually take limits as something to guarantee the highest level of detail within, so that a plywood plate actually looks like one and not resembles a sheet of paper, but that is clearly not your approach.

Every tail assembly was modelled milimeter accurate and now are just one pixel thick and invisible from a head-on/tail-on view.

But that is fine with me, as long as people are aware that that is the way you want things and not the level I could produce.

If the eMail reads "...file for the Type 91 torpedos LOD0", why would anyone consider it to be anything more?

[MicroWave]

Just one more info on item 7). Save TGA with RLE compression turned OFF. The game can't handle RLE compression.

[Zorin]

Originally Posted by MicroWave

Just one more info on item 7). Save TGA with RLE compression turned OFF. The game can't handle RLE compression.

Noted.

[EJGr.Ost_Caspar]

We will see, what we can do, regarding the samples.

Originally Posted by Zorin

Well, I usually take limits as something to guarantee the highest level of detail within, so that a plywood plate actually looks like one and not resembles a sheet of paper, but that is clearly not your approach.

Every tail assembly was modelled milimeter accurate and now are just one pixel thick and invisible from a head-on/tail-on view.

What you forget is, that:
1. it mostly won't be noticed by most of the players due to moving angles, limited zoom, cockpit always on, uninterest, unknowlede etc. or whatever
2. its look has to fit with the rest, so it doesn't break the 'visual composition' of the game

Both issues suggest the use of compromises between details and saving polys/textures, even if the limits are not reached. Thats part of game development.
If the world was perfect, there would exist quite a few more different limits for different objects or it should be defined exclusively for every new project.
But its not that way, because its easier as it is ... BUT on the other side it demands some kind of good sense for what is reasonable and what is exaggerated.

But that is fine with me, as long as people are aware that that is the way you want things and not the level I could produce.

Ha, I doubt, anyone will blame on you.

If the eMail reads "...file for the Type 91 torpedos LOD0", why would anyone consider it to be anything more?

Not everyone of the team reads the inbox. The model was handed over without the note. That issue is clear now.

[JV44Priller]

Originally Posted by EJGr.Ost_Caspar

Not everyone of the team reads the inbox. The model was handed over without the note. That issue is clear now.

Im still waiting on a response from an email I send some time ago.

[SaQSoN]

Originally Posted by Zorin

How am I supposed to know that?

Well, generally, you don't. Now you do. On a side note, however, it is rather logical to have a front projection of your model in the front view window of the Max. But, don't worry, it's a typical error, I did that too, when I was only starting.

Originally Posted by Zorin

I always paint and setup textures in a higher resolution. It was just the base setup and had no detailing yet. Of course it would be resized for the finished product...

For an objects with relatively small pixel rate (a number of texture pixels per polygon) I wouldn't recommend such approach. Because it is quite possible that after resizing the texture, a details painted on it, would shift way too much relatively to the existing UWs. And some pieces previously carefully aligned on UW seams or near 3D parts would become misaligned.
Therefore, it is strongly recommended to painted a texture exactly into the size, which would be used on the final product.

Originally Posted by Zorin

I UVW to reduce possible distortion even if that requires a little more time for proper texture painting.

It is a very good approach for a large objects and large textures. In current situation, however, with relatively small pixel rate it is better to avoid making too many seams because with only few pixels per polygon it sometimes would be impossible to align texture on the seams properly.

Originally Posted by Zorin

Besides, anything that would have had seams you removed anyway...

The main hull wasn't removed, or so it seems and it was the place with the most unpleasant number of seams in most unpleasant place (were fin attachments were supposed to be painted.

Originally Posted by Zorin

Why should I do it if I can keep higher detail within the tri-level limit I have been given?

For a starters, the very thick fins, you modeled, wouldn't be accurate to the real thing. The metal fins on the torpedo had "hydrofoil" shape, so their outer edges were almost knife-sharp. The thickest part of the fin would be at the middle of it's root and it would be around 10mm or so, judging from the photos. The only image of the torpedo with that thick fins is from USS Intrepid museum, where an inaccurate wooden replica is displayed.

Another reason is that on all other IL-2 models (planes, vehicles, ships) parts, which are thinner, then 30mm or so, are considered as 0-thin. We have to keep general visual detail of all models on approximately the same level.

And finally, making those parts 0-thin allows use of alpha channel, which allows to further reduce polygon count.

Originally Posted by Zorin

I also was told that I would not have to build for 2-sided material, but could use flipped faces and only count one side per part. Could you guys please make your mind up....

Don't see any contradiction here. You can use both approaches, even on the same model, if you please. But generally, it's easier to use 2-sided material.

Originally Posted by Zorin

Alpha channels are just to be avoided as far as I am concerned as they only allow for very crude results on this resolution levels.

Do they really? Looking on the model which was sent back to you, I can not say, they do.

Originally Posted by Zorin

Surprise, see point one. If I don't get proper info on that, how can I include it? Besides this was clearly labelled as a LOD0 preview...

LODs were mentioned along with basic triangle and texture requirements somewhere on this forum.

Originally Posted by Zorin

Also, the variants are meant to allow every type of torpedo plane to use the proper torpedo tail for any given time frame. If you don't want the four, which could easily be 8 btw, then just leave out the tail assemblies altogether, cause anything else would be a halfarsed job.

Sometimes, you got to get into account, that you work with other people, not just alone. In this particular case, making a gazilion variants of basically the same weapon, which differs from others only extrnally generates a lot more work for a programmers who will implement this ordnance into the game. Throwing at them more work at a time, when they are already overwhelmed with other work isn't a good example of team work.

That's approximately the same story, as with polygon limit, as Caspar explained it to you.

[Zorin]

Originally Posted by SaQSoN

Well, generally, you don't. Now you do. On a side note, however, it is rather logical to have a front projection of your model in the front view window of the Max. But, don't worry, it's a typical error, I did that too, when I was only starting.

Noted.

Originally Posted by SaQSoN

For an objects with relatively small pixel rate (a number of texture pixels per polygon) I wouldn't recommend such approach. Because it is quite possible that after resizing the texture, a details painted on it, would shift way too much relatively to the existing UWs. And some pieces previously carefully aligned on UW seams or near 3D parts would become misaligned.
Therefore, it is strongly recommended to painted a texture exactly into the size, which would be used on the final product.

I am obviously not used to low poly modelling. I shall adapt.

Originally Posted by SaQSoN

It is a very good approach for a large objects and large textures. In current situation, however, with relatively small pixel rate it is better to avoid making too many seams because with only few pixels per polygon it sometimes would be impossible to align texture on the seams properly.

The main hull wasn't removed, or so it seems and it was the place with the most unpleasant number of seams in most unpleasant place (were fin attachments were supposed to be painted.

Noted and I am eager to see how you guys mapped it now.

Originally Posted by SaQSoN

For a starters, the very thick fins, you modeled, wouldn't be accurate to the real thing. The metal fins on the torpedo had "hydrofoil" shape, so their outer edges were almost knife-sharp. The thickest part of the fin would be at the middle of it's root and it would be around 10mm or so, judging from the photos. The only image of the torpedo with that thick fins is from USS Intrepid museum, where an inaccurate wooden replica is displayed.

I mostly used tech drawings and photographs of the salvaged original torpedo, though my latest version of the tail planes had little to no resemblance to the original as I had to cut down polys left, right and center.

Originally Posted by SaQSoN

Another reason is that on all other IL-2 models (planes, vehicles, ships) parts, which are thinner, then 30mm or so, are considered as 0-thin. We have to keep general visual detail of all models on approximately the same level.

Good to know, then nothing will need to be made accurate in the thickness regard. Should save me some time calculating measurements

Originally Posted by SaQSoN

And finally, making those parts 0-thin allows use of alpha channel, which allows to further reduce polygon count.

Don't see any contradiction here. You can use both approaches, even on the same model, if you please. But generally, it's easier to use 2-sided material.

Do they really? Looking on the model which was sent back to you, I can not say, they do.

Alpha channels always feel like a short cut to me and that is why I have avoided them so far, perhaps you guys can teach me how small resolution textures acan produce accurate alpha results.

Originally Posted by SaQSoN

LODs were mentioned along with basic triangle and texture requirements somewhere on this forum.

Can't recall ever having seen any specifics about them in all my years around here.

Originally Posted by SaQSoN

Sometimes, you got to get into account, that you work with other people, not just alone. In this particular case, making a gazilion variants of basically the same weapon, which differs from others only extrnally generates a lot more work for a programmers who will implement this ordnance into the game. Throwing at them more work at a time, when they are already overwhelmed with other work isn't a good example of team work.

That's approximately the same story, as with polygon limit, as Caspar explained it to you.

See, how are we outsiders supposed to know that? Basic logic would dictate that a visually different model would only require a simple copy and paste effort in the coding department as nothing has to be changed except for the name and path info and that is hardly a limiting factor.

This leads us straight to an important question:

Is there a point to making new types of bombs (HE and AP) for the Japanese (and othernations) or will the bombs stay as simple as they are now and therefor it will make no difference what kind they are?

[SaQSoN]

Originally Posted by Zorin

nothing will need to be made accurate in the thickness regard. Should save me some time calculating measurements

Low-polygon modeling for games always requires to simplify RL objects for two reasons: save polygons on large objects and make smaller and simpler objects (where one could care less about polygon number) at equal detail level, so they would fit into general visual representation of the virtual world one is building.

Because, if you have too detailed object in relatively low detailed world, it would make look the whole world unrealistic, the same as if you have low detailed object in the high detailed world, it would make this object unrealistic and out of place.

Originally Posted by Zorin

Alpha channels always feel like a short cut to me and that is why I have avoided them so far, perhaps you guys can teach me how small resolution textures acan produce accurate alpha results.

IL-2 game models use alpha channels everywhere, on every plane and cockpit model there are plenty of it. Many ground objects, all ships, etc. use a lots of alpha-channel textures. Many polygon-saving techniques, used in the game models, are based on alpha-channel usage. So far, that doesn't seem a problem.

Originally Posted by Zorin

See, how are we outsiders supposed to know that? Basic logic would dictate that a visually different model would only require a simple copy and paste effort in the coding department as nothing has to be changed except for the name and path info and that is hardly a limiting factor.

In case of multiple variants of the same thing, a programer would have to program appearance of respective model according to correct service time, which would require him to check all those dates, check what model should appear when, etc. Not a big deal, when he has only this stuff to do, but another fraction of an ass-pain, if he has a lot more stuff to do as well, which should have been done yesterday and so on. So, whenever you can ease his job - why not?

Originally Posted by Zorin

This leads us straight to an important question:

Is there a point to making new types of bombs (HE and AP) for the Japanese (and othernations) or will the bombs stay as simple as they are now and therefor it will make no difference what kind they are?

I didn't say, there is no need to replace bad, or innacurate models, but I don't see much use in making a multiple variants of the same thing, which differ only visually, with no functional difference. So, in general, it is enough to have like maximum 3 visual variants of the same ordnance, as it does not affect the gameplay much. The smaller and the more "secondary" an object is - the fewer variants and less visual accuracy can it have. It's just a matter of saving labor resources.