Hard Light Productions Forums
Community Projects => The FreeSpace Upgrade Project => Topic started by: rscaper1070 on October 08, 2011, 08:16:58 pm
-
Here is what I have so far.
(http://img802.imageshack.us/img802/5996/alastorwip01.jpg)
(http://img338.imageshack.us/img338/8246/alastorwip02.jpg)
-
Stuff like this is tricky, since often normal maps represent a more efficient way to get your detail in on scales this small. Personally, I'd increase the number of sides on that cylindrical bit, and make the actual turret submodel more detailed, but save a lot of the notches and terraced bits for the normal map - possibly generated from the existing model?
-
Wonderfully detailed, definitely sticks with the presentation of the texture.
Yours seems to be a right angle squared. Which I don't personally have a problem with, but the (Retail) model seems to be more of a non-square X shape rather than a + shape.
I'd maybe suggest, as there are supposed to be two weapons platforms, make another one that is slightly more rectangular/X shaped and we can use one as the Cerberus.
I also hope the central thickness is the same too. I can also agree with BW on the maybe detailing the turret a little bit more, but I'm definitely interested in knowing the current poly count. I'd also say don't worry just yet about removing any of the currently modeled details until it's UV'ing time (which will probably be after in-game testing time).
Are you looking to UV this yourself?
-
I'd maybe suggest, as there are supposed to be two weapons platforms, make another one that is slightly more rectangular/X shaped and we can use one as the Cerberus.
Um... sorry, but the Alastor has no alternate version. The Cerberus is the better armed, identical brother of the Watch Dog, both of which are from FS1 and bare no (close) resemblance to the Alastor.
-
Really? Did I derp that hard? Okay then, nevermind. I f'ing hate cold and flu season.
So yeah, the turret could use some touching up. Something maybe like the turret on the nuMedusa. And so long as it has the same general thickness, I'm happy with the progress.
So, did anybody ever do the Watchdog/Cerberus? I know ours looks different than the retail one in some aspects, but it could use an overhaul as well.
-
I think to the exception of the guns themselves the Watchdog / Cerberus is still the Retail model.
-
Really? Did I derp that hard? Okay then, nevermind. I f'ing hate cold and flu season.
No biggie. We all brain-fart every now and then.
So, did anybody ever do the Watchdog/Cerberus? I know ours looks different than the retail one in some aspects, but it could use an overhaul as well.
I recall that someone began working on one a few years back, but it never materialized. I don't recall who was making it, but I believe that they dropped out of the community after a short while. The only upgrade that our current one has over retail is the textures.
-
Well perhaps after this one is complete.. rscaper might want to take a go at those two? :)
-
Here's a little more work on the turret. I got the central thickness correct but I'm still kicking myself for not noticing the middle was a diamond not a square. I was just eyeballing it from a movie I made from the lab. Maybe I'll try a new model once this is done and see which looks better.
I am going to try and UV this myself even though I'm still learning the ropes. So it could take awhile.
Edit: Oh yeah, right now the poly count is around 2.3k.
(http://img257.imageshack.us/img257/404/wipalastorturrent.jpg)
-
I think you should understand a bit better what a gun looks like when situated in a turret. The rotational point would be within the slotted section, not on it. The gun must not only exist within the turret housing, but have enough space to, in many cases, rotate beneath the housing and into the hull of the vessel it's mounted to. Try to find the cross section of a modern warship's turret sometime if you can. For more immediate results:
(http://img214.imageshack.us/img214/5930/5inch.jpg)
-
I think to the exception of the guns themselves the Watchdog / Cerberus is still the Retail model.
Nope, both textures and the model have been updated, though textures could use some work. The mesh is badly smoothed, but it's slightly better than retail.
-
I think to the exception of the guns themselves the Watchdog / Cerberus is still the Retail model.
Wrong, both textures and the model have been updated, though textures could use some work. The model is badly smoothed, but it's slightly better than retail.
Why do I hear this (http://www.youtube.com/watch?v=McAeQiLmEYU) tone when reading this post? Pro advice: Tact = win.
In other news.. Thaeris makes a decent point about the turret housing. Another thing to look into is if that turret can/should rotate. (Can't remember) If it should, you might consider making the base (the platform under the turret housing) more round?
All in all, though, this is progressing nicely. Keep it up. If you need any tips/help with UVing (or anything else for that matter), drop into scp-fsu IRC and we'll help you out.
-
Should be better now.
As for Alastor, it's looking great so far.
-
Right now it feels like a low poly model, made with lots of polygons. I find that what really sets a good high poly apart from a retail is the overall shape of the model. It can't just be the original silhouette with bits and pieces tacked on to increase the poly count. It has to be fundamentally more complex.
For instance, long sharp edges and points. Those are almost exclusively found in low poly models. In the your case, the Alastor has four very flat, unshaped panels meeting up with its reflection on the bottom. In the low poly model, this was because each panel was essentially one polygon. In your version yours doesn't have to stay with that simple shape. In fact it's vital that you alter shape. You need to prove to the audience that it is no longer a low poly model, and it has taken a big step away from simple shapes and a big step towards realism.
The little tidbits stuck on like cherries on a cake can be made with a normal map.
-
Right now it feels like a low poly model, made with lots of polygons. I find that what really sets a good high poly apart from a retail is the overall shape of the model. It can't just be the original silhouette with bits and pieces tacked on to increase the poly count. It has to be fundamentally more complex.
For instance, long sharp edges and points. Those are almost exclusively found in low poly models. In the your case, the Alastor has four very flat, unshaped panels meeting up with its reflection on the bottom. In the low poly model, this was because each panel was essentially one polygon. In your version yours doesn't have to stay with that simple shape. In fact it's vital that you alter shape. You need to prove to the audience that it is no longer a low poly model, and it has taken a big step away from simple shapes and a big step towards realism.
The little tidbits stuck on like cherries on a cake can be made with a normal map.
I'm sorry, but this post was way too critical and negative considering the quality of work here. If I were a new modeller, I'd probably just give up at this point. On a more subjective note, I find your criticism baseless when comparing to the retail mesh. (See below)
This is a simple gun platform for one thing. The suggestion that it must "prove that it is no longer a low poly model" has easily been met. Also, all the little tidbits can be done with a normal map... but with a lot less realism. I tend to see these things up close all the time for one reason or another. The modelled details are a nice touch. Normal maps for that kind of detail usually don't sell as well at super close range (and at low angles).
Furthermore.. "In the your case, the Alastor has four very flat, unshaped panels meeting up with its reflection on the bottom" I've tried multiple times to find the geometry on the new mesh that you are trying to critique here...
Also, I found this post to be generally condescending with a lot of generic "here's how to model" advice as if your philosophy is best. I would like to encourage more specific suggestions for improving the mesh in question and less "I find that what really sets a good high poly apart from a retail is the overall shape of the model" and "It has to be fundamentally more complex" stuff. The sentence "In fact it's vital that you alter shape" is straight up untrue and up to the modeller. In fact, major changes to the shape start to affect mission balance which is something we consider in FSU. It does it less so with smaller models, but exponentially more with the larger ones.
Finally, I am more and more adamant that we need to see a higher quality of posts in these critique threads. I've pointed out a few example posts in the past that did it well. (http://www.hard-light.net/forums/index.php?topic=64737.msg1551227#msg1551227) A simple guideline: If everything about your critique is negative, then you are probably doing it wrong.
[attachment deleted by ninja]
-
A little more progress. I tried out BW's idea but I felt it looked really bulbous and weird so I cut out the center part that I did like and put that in. I'm not sure what I'm going to do with the rest of the body yet. I'm guessing it's supposed to be some kind of station keeping thrusters so I was thinking maybe a triple head thruster nozzle or something like that.
(http://img98.imageshack.us/img98/3467/gtsgalastorwip03.jpg)
-
Looking good. I like it. Aside from trying to slot the barrels a bit more on the turret itself, there really isn't much left that I can think of to be done here.
Though, I do have some suggestions. They don't carry any particular weight to them, so consider them as me "thinking out loud" here, but here they are:
(http://img265.imageshack.us/img265/4363/alastorwipsgstn.png)
1: Flip these two support struts around, but leave the other two. This provides a semblance of architectural cross-bracing.
A very minor variation to its detailing that I think might add a nice touch to it.
2: These are just fine being flat, but I wonder what they might look like if they were beveled in, or curve-extruded.
The beveled would be cheaper poly wise, and I don't think beveled OUT would be a good idea.
Further idea: maybe only bevel in the ones that are not marked by suggestion 1?
3: An octagon "base" or surrounding plate/ring might help here and then you can drop the form fitting surrounding at the base of the turret itself, this may help it look more "natural".
-
It seems to me also that this model, as previously-mentioned, uses 90 degree angles to separate the 4 sides, but the original is supposed to have 2 acute angles and 2 obtuse angles. You may have done this, but the viewing angles make them look like all right angles.
-
Right now it feels like a low poly model, made with lots of polygons. I find that what really sets a good high poly apart from a retail is the overall shape of the model. It can't just be the original silhouette with bits and pieces tacked on to increase the poly count. It has to be fundamentally more complex.
For instance, long sharp edges and points. Those are almost exclusively found in low poly models. In the your case, the Alastor has four very flat, unshaped panels meeting up with its reflection on the bottom. In the low poly model, this was because each panel was essentially one polygon. In your version yours doesn't have to stay with that simple shape. In fact it's vital that you alter shape. You need to prove to the audience that it is no longer a low poly model, and it has taken a big step away from simple shapes and a big step towards realism.
The little tidbits stuck on like cherries on a cake can be made with a normal map.
I'm sorry, but this post was way too critical and negative considering the quality of work here. If I were a new modeller, I'd probably just give up at this point. On a more subjective note, I find your criticism baseless when comparing to the retail mesh. (See below)
This is a simple gun platform for one thing. The suggestion that it must "prove that it is no longer a low poly model" has easily been met. Also, all the little tidbits can be done with a normal map... but with a lot less realism. I tend to see these things up close all the time for one reason or another. The modelled details are a nice touch. Normal maps for that kind of detail usually don't sell as well at super close range (and at low angles).
Furthermore.. "In the your case, the Alastor has four very flat, unshaped panels meeting up with its reflection on the bottom" I've tried multiple times to find the geometry on the new mesh that you are trying to critique here...
Also, I found this post to be generally condescending with a lot of generic "here's how to model" advice as if your philosophy is best. I would like to encourage more specific suggestions for improving the mesh in question and less "I find that what really sets a good high poly apart from a retail is the overall shape of the model" and "It has to be fundamentally more complex" stuff. The sentence "In fact it's vital that you alter shape" is straight up untrue and up to the modeller. In fact, major changes to the shape start to affect mission balance which is something we consider in FSU. It does it less so with smaller models, but exponentially more with the larger ones.
Finally, I am more and more adamant that we need to see a higher quality of posts in these critique threads. I've pointed out a few example posts in the past that did it well. (http://www.hard-light.net/forums/index.php?topic=64737.msg1551227#msg1551227) A simple guideline: If everything about your critique is negative, then you are probably doing it wrong.
Obviously counting isn't everyone's strong suit :) Here, let me illustrate:
(http://i1211.photobucket.com/albums/cc433/bobbtmann/Various/ImageBin.jpg)
Heh, there's isn't much on the alastor, so you must not have looked very hard. As for the rest of my comments, rscaper1070 seems to know how to model. So I critique the form and style, and not the particulars of the geometry like triangles. When I say you have to prove that it's high poly, I mean that the overall mesh needs to be different enough that a viewer can tell right away. Not once they get up close and come to a stand still.
It would look something like this (but this is just a quick idea, it can really be anything):
(http://i1211.photobucket.com/albums/cc433/bobbtmann/Various/gtsgalastorwip03.jpg)
And hey, it's your party rscaper1070 , so model how you want to. But please consider what I'm suggesting.
-
^Looks very protoss. Also, do you own a tablet? Because that's some decent drawing.
I've always sort of imagined the Alastor to fold up somehow when not deployed. The four arms on the side look like they could fold in to leave a box-like shape conducive to shoving in some storage compartment on a corvette or destroyer. If that's sort of the impression that you have too, then a couple hinges here and there could make for some nice decoration.
-
I don't like your "revised" version of rscaper's model at all, it looks organic like something made of yorik coral rather than the mass-produced metal object that rscaper's is. It's not supposed to be sculpture, it's a gun platform. I think it's fine the way it is.
-
Don't change the shape of the FS models! Rscaper's model is awesome as it as now. Keep up good work :yes:.
-
Just make sure to get the weird shape of the Alastor in top-down view right. Retail model isn't symmetrical, looking more like some sort of fan.
-
Some more progress. I tried out my triple head thruster idea and it didn't look right at all so I went with the single nozzle.
@Zacam: I'm not sure I understand your first note. I was planning on adding some braces to the inside there anyway.
(http://img80.imageshack.us/img80/7784/gtsgalastorwip04.jpg)
-
NICE work there.
As for point 1: I was meaning that current, all four of the "braces" run like so: _/---\_ (ish). I was suggesting picking two of them (in a balanced/opposing manner) and making them go -\__/- (ish)
-
Here's what I was thinking of for the brace structure...
(http://img694.imageshack.us/img694/5797/gtsgalastorwip06.jpg)
(http://img703.imageshack.us/img703/2977/gtsgalastorwip07.jpg)
And the center is now a diamond not a square...
(http://img849.imageshack.us/img849/5492/gtsgalastorwip08.jpg)
-
I like it more now.
-
Looks wise, it's great. Feels like the old one, yet looks appropriate for a more modern engine. My only concern lies in the poly count, but without those numbers in front of me, i'll with hold any judgement on that.
-
The poly count is 3.5k right now. That's probably to many for something like this, huh? :confused:
-
Trust the power of LODs. With something of that size, you won't have to render LOD0 very often, nor in any large numbers. 3.5k sounds fine.
-
Looking fantastic. Do you think pipes/wires/computer type greebles might be appropriate in the hollows of the arms, or around the central core portion?
-
:bump:
Started texturing this one.
HTL Alastor (View in 3D (http://p3d.in/kd8XT))
(http://p3d.in/model_data/snapshot/kd8XT) (http://p3d.in/kd8XT)
-
Is there a method to screw with the model scales in FRED? 'cause that thing would work as an awesome spacestation with four big-ass cannons :)
-
You can apply any transform from any recent version of PCS2.
-
Is there a method to screw with the model scales in FRED? 'cause that thing would work as an awesome spacestation with four big-ass cannons :)
I suspect that when it's done, the size of things like grunge, lights, and other details will look rather odd if scaled up that much.
-
Very nice! Something looked off until I realized there was no AO bake (inb4 Kolgena)
I especially like the look of the turrets here. Can't wait to blow it up :D
-
Yeah... except there IS an AO bake.
-
I swear it looked different this morning. I blame the fact that I had just woken up. :P
-
And here is the final look.
(http://i282.photobucket.com/albums/kk264/mjnmixael/Private/screen0434_zps6a158ed7.jpg)
(http://i282.photobucket.com/albums/kk264/mjnmixael/Private/screen0435_zps50b302e0.jpg)
(http://i282.photobucket.com/albums/kk264/mjnmixael/Private/screen0436_zpsf48516bb.jpg)
-
Yet another in a long list of excellent upgrades. Are we going to see this released standalone or will it be held until the next mediavps?
-
Yeah... except there IS an AO bake.
I'm not a graphics guy, but I tend to associate "AO bake" with a bit of darkness along a concave edge. Why is that not the case here?
-
Oh. My. Gawd.
Do you guys get off on AO Bakes? Lookin for some AO Porn? Sheesh.
You need frickin proof that it's there? Apparently so... well..
Here.
(http://img259.imageshack.us/img259/2498/imagebina.jpg)
It.
(http://img809.imageshack.us/img809/9918/imagebin.jpg)
Is.
(http://img715.imageshack.us/img715/9918/imagebin.jpg)
I have half a mind to stop adding AO bakes just to be a dick since you guys seem to think an AO Bake is the only thing that can make a texture job OMFG AWSUM. :doubt:
-
Don't listen to them mjn.mixael, I only care about the diffuse layer! ;)
-
Oh. My. Gawd.
Do you guys get off on AO Bakes? Lookin for some AO Porn? Sheesh.
Oh baby. That gets me so hot. Oh yes. Yes.
-
Do you guys get off on AO Bakes? Lookin for some AO Porn? Sheesh.
All porn is AO porn. :p
As for the texture job, it looks good, but I think the dirt or noise or whatever in the shinemap might be a bit too strong. Where the light hits it ont he top half of this image (http://i282.photobucket.com/albums/kk264/mjnmixael/Private/screen0434_zps6a158ed7.jpg) it looks like it's been bitumized or something.
-
You need frickin proof that it's there? Apparently so... well..
I am sorry to frustrate you, but you have misunderstood me completely. I only thought about it because someone else questioned it, and I was just hoping someone more experienced than myself could provide an explanation.
I was honestly curious, and still am curious, why it wouldn't look more like this:
EDIT: I'm not even sure it should look like that, it's just that it's what I tend to see in examples of AO. Perhaps it was a flaw in early techniques that I'm not familiar with?
[attachment deleted by ninja]
-
Yeah, there's something... odd... about the way you did that AO.
Is that per-vertex instead of per-texel?
-
Hmm, I may have found what I'm looking for: http://nothings.org/gamedev/ssao/ (http://nothings.org/gamedev/ssao/)
No time to read it now unfortunately, I'm up quite late as it is.
-
Yeah, there's something... odd... about the way you did that AO.
Is that per-vertex instead of per-texel?
We have an AO expert here! Aardwolf will now be diong the AO for all my future texture jobs.
Hmm, I may have found what I'm looking for: http://nothings.org/gamedev/ssao/ (http://nothings.org/gamedev/ssao/)
No time to read it now unfortunately, I'm up quite late as it is.
If I'm reading that article correctly, he would like my AO bake.
-
I think I would rather have a model with no AO bake than no new model at all. It having an AO bake is just a bonus for me.
And I really don't see how this AO bake could be considered inferior or unusual. It just looks like the recessed place is shadowed, and you know what? That makes visual sense.
-
If I'm reading that article correctly, he would like my AO bake.
Indeed. I've misplaced the link, but I found another article earlier today that indicated that baking the dark corners into diffuse is a common mistake, and they should only be used when you have the option to only have them affect ambient light, and not point sources. Like most lighting hacks it's not supposed to be realistic, it's just better than having no shadows at all in an area with little or no direct light.
-
Ironically, deep space has essentially 0 ambient light. Only the FS2 nebula, atmosphere missions, and heavy firefight set pieces could physically justify an AO bake. At least those are the same scenarios where the diffuse would be bright enough to see the AO at all (unless you don't use dark as hell lighting settings, in which case realism isn't a concern anyways)
I still think it looks better, especially since shadow mapping is still by and large really expensive.
(At risk of being a hypocrite, I think we should probably stop talking about AO theory... this is mjn's Alastor thread after all)