nybt-forum-threads PBGV's BAT Thread

[PBGV103 729]

In this render I changed a few of the blinds on the front of the building. I do like the more organized look, especially from the more distant zooms. It took a little while, but the most time consuming part was just deciding which groups to change. The blinds are actually just 18 separate objects, so changing them in small groups is fairly easy. It's not as though I actually need to change each one individually.

Francis, with this model I'm trying to make it look as it is currently, so I will be keeping the pull down blinds. For the area with curtains I'll probably end up adding those as well. Since I have the same type of curtains in 299 Park Ave and 345 Park Ave I can just import them into the scene.

[Jasoncw 11,949]

Yeah I think that looks much better.

[gutterclub 2,729]

Yep, something so simple can make such a big difference. Love it

[darn42 1,268]

Btw, I think you've hit a great balance with the windows. Their realism now matches the quality of the rest of the building.

[Roger51 381]

Stunning!! That is amazing, can't wait to have this in my cities!!

[Francis90b 748]

That's a noticeable improvement .

[libellibicus 396]

I'm really looking forward to seeing this in my cities

[Darknono35 3,138]

Perfect for my downtowns, it's a great improvement !

[kellydale2003 2,276]

I know I go by 133% since its mathematically correct, but to me it almost looks like it wasn't scaled up at all. Maybe my eyes are playing tricks on me though.

Mathematically correct is 100% :b

But the illusion of mathematically correct in Simcity 4's Orthographical views is supposedly 133%.

Personally I've done my own 1x1x1m test which resulted in a 12x10 pixel face of the box. I did a 1x1x1.2m box and found the pixel face of the box to be 12x12... But there was this transition pixel that is kind of the middle guy. It was half grey and half green... which in theory COULD be the 133% correct scale to the orthographic view... but when I did a 1x1x1.33m box, the transition pixel was heightened by 1 pixel... which if in theory is the correction scale for the orthographic views would only have 12x13 pixel resolution rather than a 12x14 pixel face (if the transition pixel was considered to be part of the building).

So in my personal opinion... 120% is correct. Sure sure we might have a huge debate on this again, but that's my opinion. If you scale the 1x1x1 box by 1000% the factor of scale is increased, but not the pixels themselves. This is why I kept it at a 1x1x1m box instead of something significantly higher (because essentially working in the smallest of numbers for an equation is much more simple than working on a large scale. Anyone would know that in precal or calculus math! haha... the smallest of uncalculated digits can have a huge fogging impact on your final result.)

Some people might think that 125% might also be correct... it makes much more sense to work in quads than any other, but I can assure you it's been an ancient question since the Babylonians! I've never tested 125%. but it's pretty relative and reconciled to the divided 120% and 133%.

Others just like to BAT in 100%... but I personally choose that the illusion of correct height is better than the illusion of incorrect height.

But anyways... for your glass, it looks alright... but the reflection of the plane could be influencing your glass more than anything. When I have good correct glass settings that work for another building, but not for a current project.. I play up the reflecting plane to help fix or compensate for the problem.

As for the blinds, I don't have much problem with them either.

As for the empty floors everyone keeps talking about... It's Zoom 5 (a zoom level not experienced well in-game, and not one that everyone can pay attention to). Zoom 4 will take this illusion of emptyness away, but if you're looking for a high-quality solution...I know this personally that it can be time consuming. And a recent fix of mine is creating a bunch of geometry at the bottom with random textures to look similar to office furniture. Have some of them over lapping... then array the geometry by the number of floors ... then I use this random select script for 3dsmax..

Random Select Script

I then randomly select a desired number of geometry and delete it! then I freeze all other geometry, and randomly select floors and group the geometry... then I rotate it by 90 degrees, then do it again and again. Then ungroup all the geometry... selectively and randomly rotate some of the geomtry inside. Some of my chairs just have black standard material made from lines... and it looks pretty good! haha... I don't have a photo yet, nor will I say WHICH project it is that I used this technique on... but it could be a solution or partial help to what you're looking for! (:

, K

EDIT: ps; If you want a better way to do blinds... here's a script.

http://www.cgtechniques.com/goodies/blinds/

This is the one I edited for my own personal use, if you need help to edit it just PM me!

It's better than using a couple objects, and there are many parametres for a variety of effects... MUCH better than nearly a hundred objects or so! And it only takes seconds! (:

[PBGV103 729]

Kelly, you jumped to conclusions without reading my entire posts... The script you linked to is the exact same one I've been using. I never had 100s of different objects for the blinds, only a few generated by the script.

As for the 133% argument, I use it because it looks correct to me and from what I've seen, the majority of the other modeler's here who have picked an actual scale also picked 133%. I saw a write up someone did while ago where they had actually come up with some kind of formula for calculating what vertical scale would look correct in the SimCity 4 view and they came up with 4/3 original height. Of course I suppose their math could've been wrong, I wasn't paying a lot of attention, but I have to say that when I was working on the Citigroup Center's roof, the only scale that kept the slope 45 degree slope looking correct was 4/3 original height. That to me is proof enough that 133% mathematically maintains the correct look of the building.

[sten1700 102]

Both beauties! Can't wait!

[AndisArt 2,068]

oh how very pretty!

[Cockatoo 2,257]

So in my personal opinion... 120% is correct.

The problem there is that you used such a small sized box to carry out your tests the results would’ve no doubt been lost in the antialiasing. If you can muster the patience you should try again, this time with a box maybe a few hundred px wide and see what results you get. This’ll give you a better margin for error.

Mathematically 133% I think will give you accurate proportions in game. But really I don’t think scaling matters that much any more. Some do, some don’t, there’s no one system to conform to because even among those who do stretch people still model things (i.e. floor heights) a bit differently to each other.

Anyway, this looks pretty awesome PBGV, especially that base.

[Jasoncw 11,949]

Kelly it's 133% and the last time you talked about it everybody said the same thing as they are now. Your cube was too small. Or if a smaller cube is more accurate then make a cube 0.001 m and tell me how many pixels each direction is.

[PBGV103 729]

Nice demonstration Jason.

On another note, you may remember me talking about receiving a bad batch of memory from Corsair. Well the replacement finally came in two weeks after I returned the defective ones and guess what! These are defective too! I guess I'll be spending another $20 on shipping them back and waiting another 2-3 weeks for more replacements, which will probably also be bad. You would expect a little quality control from a reputable company like Corsair and for memory that costs $350. My cousin is an attorney and I may be giving her a call soon...

[Cockatoo 2,257]

Bummer about the memory. $350... very expensive, did you get 32GB of the high quality gamer stuff or something? Corsair is supposed to be one of the best memory makers... you should definitely demand a discount if they've sent you broken parts twice in a row. But are you sure it's not the motherboard first?

[PBGV103 729]

But are you sure it's not the motherboard first?

I already spent $120 having a group test every component in the computer the first time I had memory problems. That's when they narrowed it to the Corsair memory. I will be calling them in the morning and asking for three things: an advance replacement of the memory, their own testing of the memory before they send it, and shipping that ensures that it arrives by Saturday because on Sunday I have to go back to school which is 300 miles away from here.

[Jasoncw 11,949]

Yeah that's really dumb. If the part is defective they should pay for the shipping. It's understandable that that would happen sometimes (although twice in a row??) but that's a cost that's usually eaten by the business and not the buyer.

[Darknono35 3,138]

I love your Citigroup it's an important skyscraper for re-doing NYC. The details looks awesome and gorgeous

[Seraf 1,239]

You really came to mastery with those glass buildings. And like always I admire all details you put in your BATs.

[kellydale2003 2,276]

Whether pixels are lost in the antiliasing or not, it's really the illusion of correct scale that's important here (in my opinion).

My Trump Tower scaled at 133% was ridiculous! It was taller than Rubik's Twin Towers! haha... but yeah like I said, I really don't want to get into a massive debate about this again... It's kind of like arguing which religion is right! :b

[gutterclub 2,729]

Yeah, but his Twin Towers might be completely underscaled? As mentioned before, if people stick to RL heights and scale on the Z axis accordingly then all BATS will be the same scale and nobody has to work out new mesasurements or make stuff up? That said, 125% isn't a great deal difference...as long as it IS scaled up, otherwise it looks silly and squashed

[PBGV103 729]

I've never seen a building that looked "wrong" when scaled up 133% in height when the dimensions were correct to begin with, but I guess for me it's mainly a big deal because I'm not making these models just for random buildings to add to the game. Everything I'm working on is for my NYC recreation. Of course I love to use other people's buildings as well, but when everyone is using a different scale all of the buildings start not looking right together. Then I start to feel like I'll have to make my own version of those building so they'll fit it.

If you go in with the idea that you're going to make something that only resembles a real life building, then just go for it, but for me, I'm sticking with 133% because it undoubtedly works and so far all of my models along with, from what I understand, the majority of the other models this team as worked on are at that scale.

[Francis90b 748]

My Trump Tower scaled at 133% was ridiculous! It was taller than Rubik's Twin Towers! haha... but yeah like I said, I really don't want to get into a massive debate about this again... It's kind of like arguing which religion is right! :b

Unless you altered the Trump's dimensions in any way (i, for now, assume you didn't), it's quite evident that Rubik did use another scaling factor for his WTC towers and/or manipulated it's dimensions somehow...and when someone else BATed something according to dimensions and 133% scaling, it became apparent that things weren't going to match.

Someone said religion is people's opium...freaky theories about scaling and manipulating dimensions according to g-d knows what criteria might quite well be BATer's opium...

It would be all nice and good if deciding for one of them could have no nasty repercussions, but this thread's author, and i'd back his own testimony, said that this doesn't happen - to the disadvantage of final users...

As far as i know, one BAT (either a recreation or something depicting one possible reality) should reflect the following:

Mass

Details (when reasonably visible)

Proportions

Materials.

Working with some "chosen" dimensions rather than the building's original ones alters the appearence of the mass, and might in some occasions even alter building's proportions.

Working with one arbitrarily chosen scaling factor will alter building mass, if the scaling factor is about the vertical axis, will alter building's proportions and mass - not to mention that even details might be altered - look at PBGV 103's post about the 45 deg slope of the Citigroup building, or try making one wall with one semicircular arch and scale it to various amounts to see what i mean.

When all of the buildings follow the same scaling criteria, a whole new dimension is added to their overall feeling - the relationship between correctly-sized masses - to get what i mean try playing, for example, at SC3000 - which often has far more incorrect scaling relations between buildings than what you'd usually find in SC4.

it would be worth giving it some thinking, since some people spend hours of their life to decide down to the fourth decimal what's the correct setting to render one glass material.

[kellydale2003 2,276]

I like your comparison to religion vs. this conundrum! At the end of the day... it fits this theory like a glove! haha

Each religion (100% vs. 120% vs. 125% vs. 133%) all say they're correct. When it comes to far away visualization any of them can be correct... come to closer analysis and the gap is widened for which is virtually or visually correct. But at the end of the day... Each religion is definite on preference!

I guess we can all agree to disagree :b

But I still like my 120% better! :b

Anyways! I won't be posting about this topic again simply because this is Paul's thread! Sorry Paul! (:

[Francis90b 748]

You nailed one very important point with the tiny 10 px cube which, according to you, was the perfect way to find wheter 120 or 133 were exactly correct - results were, one tiny 10 px cube with one pixel blurred, and hardly one pixel difference between them.

Often, the things we look at are not clear - there's not an easy answer - many ways to interpret it are possible, this doesn't mean all of them (or even any!) of them are actually correct.

Other things, far more important than BATing, have been and are an epic mess.Religions, pretty much all of them tried to create an articulate system aimed at giving an answer - but they were often more focused into organizing what was known, both in nature and in ethics, into a coherent system (basically, organizing what's already known) than into looking for new datas and elements, and analyzing them, as far as possible, with no bias.

Sometimes religion didn't interfere with scientific process, some other times it imposed it's beliefs unto it, trying to stop or delay the acceptance of facts and elements that didn't fit into their pre-made systems.

But, while it's easy to embrace one system of belief (that, when becoming public and commonly accepted, might often deserve the title of "religion") it's also true that the scientific method has allowed us to know things that couldn't have been known otherwise....

So, while it's easy to adopt a religion, maybe picking the one most pleasing to us, is it actually worth to adopt one?

In your cube demonstration, that with the 10 (actually 13) px cube, you tried showing, with one cube so blurry that no one could actually tell what was going on, that your system was the best, and anyway, everyone could see anything into there, and that it's all a matter of personal opinion.

In this second post, you showed a sphere, one semisphere and one cube (this time, yeah, at an acceptable size) - but not even this is a "scientific" demostration - it just shows some primitives, without actually saying anything, it just gives people freedom to adopt whatever system someone considers as most fit.

Jason has tried one one demonstration with one reasonably-sized cube, and trying to pick lenghts from there - that's something far more serious.

EDIT:

This said...one experiment by mine to determine the correct scaling factor.

def:A scaling factor is one number, by which the height (z axis) of a cube rendered to SC4 perspecive should be multiplied to in order to make it equal to the cube's base (x axis).

I did the following:

- Doing a scene with one cube, placed so that one corner of it was coincident with Max's X and Y axis intersection.

- Set the viewport to SC4's perspective, as if to make a preview render.Cube was not scaled in any way.

-Take a screenshoot of the result.

-Cut and paste the screenshoot in Paint.

-Crop it (without any resizing) till it was a perfect square.

-Reopening max, making a new scene, one big square(100x100 mtrs) plane

-Pasting the screenshoot into it as a texture with the minimum blur possible.

-Drawing lines overlying the screenshoot's X axis, as well as the cube's two vertical edges starting from it.

-Since my goal was to make all sides of the cube equal, i've drawn one n-gon, 90 sides (the closest approximation to a circle we have in MAX) - centered into the screenshoot's axis intersection and with a readius equal to the cube's lenght.

-I then drew another circle, centered into the screenshoot's axis intersection, with a radius equal to the height of the "unscaled" cube.

-I compared the two radiuses - 11,941 to 8,935 - and found the ratio between them - 133,642.

-Therefore, the ratio between corrected height and unscaled height should be of 1,33642 - which means that 133% as a scaling factor should substantially be correct.

Pic:

[kellydale2003 2,276]

1) I wasn't intending to show ANY "scientific" demonstration what-so-ever!

2) I never blurred ANY of my photos! An attempt to discredit me for a discrepency is ridiculous!

3) And I'll definitely BAT my BATs the way I want, not because somebody else told me so.

4) "Since my goal was to make all sides of the cube equal, i've drawn" <<-- Definitely a bit subjective in an experiement that can pollute the end result.

5) My photo I posted was the same objects at 10x10x10m with the left being 100% scaled vertically, the centre at 120%, and the right at 133%. There was nothing biased making this photo... although we can all agree that the full sphere looks best at 100%. The 133% (in my opinion) looks far to stretched for each and every of the geometry!

6) Circles are not good in compensating for orthographic views!

7) I don't appreciate starting a dog fight in this thread! If you'd like, I'd discuss it in another... perhaps if you feel like continuuing this debate, you can make one! Other than that, I'd respect Paul and leave this thread for his showcase instead of a debate mainly based on preference and "scientific" data.

[Francis90b 748]

Much noise about nothing.

1) I wasn't intending to show ANY "scientific" demonstration what-so-ever!

Any point, and from where that point comes from, might be discussed - it's in your own interest to make sure it has some ground - wheter you'd like to call it a scientific demonstration or not.

2) I never blurred ANY of my photos! An attempt to discredit me for a discrepency is ridiculous!

You didn't.You just made an experiment aimed at finding one difference with an object that was too small to be an appropriate testing material - That's what i wrote, and i am responsible for what i wrote - not for the way you "interpreted" it, about which, honestly, i don't care.

3) And I'll definitely BAT my BATs the way I want, not because somebody else told me so.

Again, whaat? Where does it come from?

4) "Since my goal was to make all sides of the cube equal, i've drawn" <<-- Definitely a bit subjective in an experiement that can pollute the end result.

Not really - that's exactly what everyone using a scaling factor (120, 125, 133, etc) is trying to get.

To be fully honest, SC4's perspective is actually a trimetric projection, this means, each side of an object would actually need a different foreshortening coefficient on each side.Of course that's not possible, therefore i limited myself to the relation between X and Z axis.

I simply did so with one method, rather than relying on "the look" of it".

5) My photo I posted was the same objects at 10x10x10m with the left being 100% scaled vertically, the centre at 120%, and the right at 133%. There was nothing biased making this photo... although we can all agree that the full sphere looks best at 100%. The 133% (in my opinion) looks far to stretched for each and every of the geometry!

If anything, to many people in the past centuries it looked like the sun revolved around the earth, then someone showed them it didn't exactly work that way...

6) Circles are not good in compensating for orthographic views!

If i want the two sides to be of an equal lenght, the best way is to draw one circle with them as a radius - it's as simple as this.

7) I don't appreciate starting a dog fight in this thread! If you'd like, I'd discuss it in another... perhaps if you feel like continuuing this debate, you can make one! Other than that, I'd respect Paul and leave this thread for his showcase instead of a debate mainly based on preference and "scientific" data.

The expression "dog fight" it's quite wrong - it's not about you, it's about the points you make - it's called "discussion"

Any discussion, in any thread, about any subject that might be related to the subject of the thread (therefore, even vertical scaling on BATs) is relevant.

I'd say that it's up to Paul and to the Mods to decide wheter this specific discussion can stand there, or should be placed elsewhere - not on you - also, it's quite odd that this very same discussion was seemed a very fitting contribution to this thread when you started it, huh?

[kellydale2003 2,276]

At this point I refuse to discuss more about this topic on this thread... out of politeness of course. I already messaged Paul on Facebook about this.

And I'd rather not debate on the point "who started what"... that's just plain immature. Starting a new thread on this topic is 100% at your disposal. (:

[Francis90b 748]

I'd like to partecipate in a discussion about this topic, at any place where it should take place - including this one, if necessary.

Opening a specific thread would probably not do - for this topic, like many others, would still resurface here and there - therefore, it would fail to fullfill it's very reason of existence.

Personally, I would like to hear your answer about the last points i've made - if you feel like that, let's do it in private - you can always drop me a line.

And if, then, you think that some mind-blowing truth has emerged from this eventual private confrontation, and public should be enlightened about it, you can make a thread out of it - is 100% at your disposal (: .