31 replies to this topic

[lordkrike]

I have been playing with the python imaging library and decided to write a script to estimate the cross-sectional area of the 'mechs in the game. Using the images from Adridos's work, I counted up the numbers of pixels and did some conversions to get a rough estimate.

The results are here or here. Remember that these are estimates, since my measurements are only as accurate as the image profiles available. Additionally, I counted grey squares as 1/2 of a pixel.

The only interesting result I found was that three of the more fragile chassis - the Dragon, the Catapult and the Centurion - all had very low tonnage to front cross-sectional area. Hitboxes may or may not line up accurately with these profiles.

Edit: To be clear, the height and width measurements of the models are done at the widest points, which, due to posing makes some 'mechs seem unnaturally wide. That measurement was not used to compute the cross sectional area.

Edit2: adding a graph: cross sectional area per ton; a lower bar is better:

Edited by lordkrike, 14 January 2013 - 06:18 PM.

[Team Leader]

Hmmm. Good work. But are you sure those size comparisons are accurate? I thought they were just samples

[Asmudius Heng]

More proof that for tonnage - the light mechs are too small compared to other mechs.

Look at the percentage jump between a raven and a hunchback, thats 15 tonnes difference. Now look 15 tons more to a dragon.

The difference between a Hunchback and a Dragon is not nearly as massive compared to the hunchback and a the raven.

Also the Centurion compared to the hunchback is pretty dire for the same tonnage.

[Noth]

I find you calculations for width misleading as they all seem to be to the widest point on the image rather to the same point over what is generally the highest mass area of the upper torso and shoulders. They way you have done it give a pretty poor indication of how wide the mechs actually are. The Centurion for example is wider than a hunchback, you simply took the extended hands into the width of the hunchback which is not accurate.

[lordkrike]

Team Leader, on 14 January 2013 - 03:24 PM, said:

Hmmm. Good work. But are you sure those size comparisons are accurate? I thought they were just samples

Unsure about that one. People in the thread were using them to compare heights.

Noth, on 14 January 2013 - 03:57 PM, said:

I find you calculations for width misleading as they all seem to be to the widest point on the image rather to the same point over what is generally the highest mass area of the upper torso and shoulders. They way you have done it give a pretty poor indication of how wide the mechs actually are. The Centurion for example is wider than a hunchback, you simply took the extended hands into the width of the hunchback which is not accurate.

As far as the straight width measurement, yes. But that's not how the cross sectional area is computed. Those arms are still hittable.

[TB Freelancer]

Asmudius Heng, on 14 January 2013 - 03:54 PM, said:

More proof that for tonnage - the light mechs are too small compared to other mechs.

Not really. You're trying to tie outer dimensions directly to weight on a 1:1 ratio but things don't work that way. Just take water as an example.

1ml would make a cube 1cm on each side and weighs roughly one gram. 1,000ml would make a cube 10cm on each side. It would only be 10 times bigger, but one thousand times heavier....

...its not really quite that simple, but I imagine that the guys who originally made the TROs for mechs didn't just pull dimensions out of the air and used existing machinery as benchmarks to come up with numbers that are plausible.

BTW...its been a long time. Hope to see you on the field sometime.

Edited by TB Freelancer, 14 January 2013 - 04:41 PM.

[lordkrike]

TB Freelancer, on 14 January 2013 - 04:40 PM, said:

Not really. You're trying to tie outer dimensions directly to weight on a 1:1 ratio but things don't work that way. Just take water as an example.

1ml would make a cube 1cm on each side and weighs roughly one gram. 1,000ml would make a cube 10cm on each side. It would only be 10 times bigger, but one thousand times heavier....

...its not really quite that simple, but I imagine that the guys who originally made the TROs for mechs didn't just pull dimensions out of the air and used existing machinery as benchmarks to come up with numbers that are plausible.

BTW...its been a long time. Hope to see you on the field sometime.

I think the more interesting part is that the Jenner and Raven have such small dimensions from the front for their tonnage.

No matter how you look at it, their frontal cross section is very small for their tonnage, even when compared to other light 'mechs, and that is definitely a kind of advantage.

Also, I'm not here to claim anything is OP or needs nerfed or is x, y, or z. I'm just here to tell you how big those 'mechs are.

[Noth]

lordkrike, on 14 January 2013 - 04:39 PM, said:

As far as the straight width measurement, yes. But that's not how the cross sectional area is computed. Those arms are still hittable.

In reality though, they don;t stick out that far in gameplay, and they are also smaller targets than the other mass heavy areas. The mass heavy areas also denote kill areas

[Wun]

You really would not expect a terribly consistent density anyway. Some mechs are narrower and deeper than others. The mechanics in some mechs will be very cramped and in some there will be lots of room for mechanics to work (no different than engine wells in automobiles and other vehicles).

[lordkrike]

Noth, on 14 January 2013 - 04:56 PM, said:

In reality though, they don;t stick out that far in gameplay, and they are also smaller targets than the other mass heavy areas. The mass heavy areas also denote kill areas

I'm trying to make the point that the width is not really the issue at hand. It's the cross sectional area, which still applies.

You make a fair point that the posing could be off (I suspect as much for the Commando, I don't seem to recall his arms holding that pose in game), but I'm not concerned with what can or cannot cause kills shots. I'm concerned with how big the 'mechs are when viewed from the front and side vs. their tonnage.

[Noth]

lordkrike, on 14 January 2013 - 05:01 PM, said:

I'm trying to make the point that the width is not really the issue at hand. It's the cross sectional area, which still applies.

You make a fair point that the posing could be off (I suspect as much for the Commando, I don't seem to recall his arms holding that pose in game), but I'm not concerned with what can or cannot cause kills shots. I'm concerned with how big the 'mechs are when viewed from the front and side vs. their tonnage.

And I'm telling you that someone reading your data would think that the Centurion is much narrower than many of the mechs when in fact it is as wide or even wider than the ones listed as having more width. It's not accurate to how big they are in real use.

[Shumabot]

TB Freelancer, on 14 January 2013 - 04:40 PM, said:

Not really. You're trying to tie outer dimensions directly to weight on a 1:1 ratio but things don't work that way. Just take water as an example.

1ml would make a cube 1cm on each side and weighs roughly one gram. 1,000ml would make a cube 10cm on each side. It would only be 10 times bigger, but one thousand times heavier....

...its not really quite that simple, but I imagine that the guys who originally made the TROs for mechs didn't just pull dimensions out of the air and used existing machinery as benchmarks to come up with numbers that are plausible.

BTW...its been a long time. Hope to see you on the field sometime.

I assume that's exactly what they did. If it isn't than whatever it was they actually did do failed miserably. Mech scale has always been wildly varying and nonsensical.

[lordkrike]

Noth, on 14 January 2013 - 05:03 PM, said:

And I'm telling you that someone reading your data would think that the Centurion is much narrower than many of the mechs when in fact it is as wide or even wider than the ones listed as having more width. It's not accurate to how big they are in real use.

I again direct your attention to the Front Cross-Sectional Area column. It shows that the Centurion is 20% larger than the Hunchback when viewed from the front.

If you have such an issue with the posing and irrelevant width measurement, we can find someone to redo the silhouettes in more natural poses.

[Asmudius Heng]

TB Freelancer, on 14 January 2013 - 04:40 PM, said:

Not really. You're trying to tie outer dimensions directly to weight on a 1:1 ratio but things don't work that way. Just take water as an example.

1ml would make a cube 1cm on each side and weighs roughly one gram. 1,000ml would make a cube 10cm on each side. It would only be 10 times bigger, but one thousand times heavier....

...its not really quite that simple, but I imagine that the guys who originally made the TROs for mechs didn't just pull dimensions out of the air and used existing machinery as benchmarks to come up with numbers that are plausible.

BTW...its been a long time. Hope to see you on the field sometime.

While reality is what reality is - my beef i think is that the overall game balance that was ported from TT and then adjusted does not take into account the size of the mech. there was no penalty in shooting at an atlas moving the same speed as a raven.

If an atlas could move over 100 in MWO it would still be far far easier to hit than the raven.

This makes light mechs very survivable as you are hard pressed to hit and hit the same location if the light knows what he is doing. An atlas will hardly have any splash damage across its sections as it is so big and slow.

It is kind of a multiplier of survivability having smaller sizes.

I wouldnt advocate changing anything until the lag is better fixed though - perhaps the biggest bugbear. However it does show they are much smaller and the size of a mech does make a big difference.

The centurion as stated is far bigger than the hunchback for the same tonnage. I find the centurion to be one of the least durable mechs and this is one big reason for it.

You may see me around - depending on how much having a child changes my gaming habits!

[MoonUnitBeta]

Is this the reason for the crazy fluctuations in data?
Taking the catapult for example, and what could be assumed though front and side profiles, the catapult's torso could look like this:

It's a crude example, but note that there isn't sufficient data in those images to tell you that "no, that's not what the missile racks look like." Anyways, that's probably the reason for the huge jumps between mechs. I do like OP's motivation though and the data he collected and compiled is interesting to look at.
Like the dragon, it's CT extends very far forward. Since there's no depth in the front view, and since there's no depth in the side view, it means that it's bicep protrude as far as it's CT.

To probably generate the most accurate data, is to take the in-game models, put them into Autocad or something, and get it to punch out the volume of each mech, and only go by height as the factor, or disect the data into groups. i.e: Atlas Arms = 10 tons each, legs = 15 tons each, etc etc. Total height: XXmeters, Total weight: XXtons

Edited by MoonUnitBeta, 14 January 2013 - 05:28 PM.

[Dirk Le Daring]

lordkrike, on 14 January 2013 - 03:14 PM, said:

I have been playing with the python imaging library and decided to write a script to estimate the cross-sectional area of the 'mechs in the game. Using the images from Adridos's work, I counted up the numbers of pixels and did some conversions to get a rough estimate.

The results are here. Remember that these are estimates, since my measurements are only as accurate as the image profiles available. Additionally, I counted grey squares as 1/2 of a pixel.

The only interesting result I found was that three of the more fragile chassis - the Dragon, the Catapult and the Centurion - all had very low tonnage to front cross-sectional area. Hitboxes may or may not line up accurately with these profiles.

Edit: To be clear, the height and width measurements of the models are done at the widest points, which, due to posing makes some 'mechs seem unnaturally wide. That measurement was not used to compute the cross sectional area.

Can you put the info up in another format please. I cannot open the file.

MoonUnitBeta, on 14 January 2013 - 05:21 PM, said:

take the in-game models, put them into Autocad or something,

Any ideas on how this may be done ? I'm clueless.

[MoonUnitBeta]

Dirk Le Daring, on 14 January 2013 - 05:26 PM, said:

Can you put the info up in another format please. I cannot open the file.


Any ideas on how this may be done ? I'm clueless.

No, I am too. Haha. I'm not sure if you can somehow extract the model out of crytek, and convert it into 3DS max where it was probably modeled.
I don't know if 3ds max can calculate volume, but it's easy enough to import 3ds max files into
autocad, and also autocad into 3ds max, however some of my models that i've tossed between 3ds max and autocad have lost some accuracy.

I don't know anything about Crytek, a little bit about 3DS Max, and more about Autocad.


I'm not going to read through this thread, but google brings up some stuff when you google "exporting crytek models into 3ds max"
http://www.moddb.com...ne-3-to-3ds-max

Edited by MoonUnitBeta, 14 January 2013 - 05:36 PM.

[lordkrike]

MoonUnitBeta, on 14 January 2013 - 05:21 PM, said:

Is this the reason for the crazy fluctuations in data?
Taking the catapult for example, and what could be assumed though front and side profiles, the catapult's torso could look like this:

It's a crude example, but note that there isn't sufficient data in those images to tell you that "no, that's not what the missile racks look like." Anyways, that's probably the reason for the huge jumps between mechs. I do like OP's motivation though and the data he collected and compiled is interesting to look at.
Like the dragon, it's CT extends very far forward. Since there's no depth in the front view, and since there's no depth in the side view, it means that it's bicep protrude as far as it's CT.

To probably generate the most accurate data, is to take the in-game models, put them into Autocad or something, and get it to punch out the volume of each mech, and only go by height as the factor, or disect the data into groups. i.e: Atlas Arms = 10 tons each, legs = 15 tons each, etc etc. Total height: XXmeters, Total weight: XXtons

I didn't actually measure the surface area of the 'mechs. I measured the surface area of their silhouettes. Another way to say it is that I measured their 2D projections and depth does not come into play there. So, when you are and an Atlas are having a squared-off staredown, you see a 'mech with 127 square meters of surface area since you're looking at a 2D projection of a 3D object. The 'mech's total surface area is actually much larger.

Your second part is completely true. That's actually where those silhouettes come from, if you look at the first link. My measurements are only as accurate as the .jpgs provided, whereas I'm sure Autocad has a function to compute the cross sectional area of an object along an axis.

[Dirk Le Daring]

MoonUnitBeta, on 14 January 2013 - 05:33 PM, said:

No, I am too. Haha. I'm not sure if you can somehow extract the model out of crytek, and convert it into 3DS max where it was probably modeled.
I don't know if 3ds max can calculate volume, but it's easy enough to import 3ds max files into
autocad, and also autocad into 3ds max, however some of my models that i've tossed between 3ds max and autocad have lost some accuracy.

I don't know anything about Crytek, a little bit about 3DS Max, and more about Autocad.


I'm not going to read through this thread, but google brings up some stuff when you google "exporting crytek models into 3ds max"
http://www.moddb.com...ne-3-to-3ds-max

No worries, thanks anyway.

[Dirk Le Daring]

lordkrike, on 14 January 2013 - 03:14 PM, said:

I have been playing with the python imaging library and decided to write a script to estimate the cross-sectional area of the 'mechs in the game. Using the images from Adridos's work, I counted up the numbers of pixels and did some conversions to get a rough estimate.

The results are here or here. Remember that these are estimates, since my measurements are only as accurate as the image profiles available. Additionally, I counted grey squares as 1/2 of a pixel.

The only interesting result I found was that three of the more fragile chassis - the Dragon, the Catapult and the Centurion - all had very low tonnage to front cross-sectional area. Hitboxes may or may not line up accurately with these profiles.

Edit: To be clear, the height and width measurements of the models are done at the widest points, which, due to posing makes some 'mechs seem unnaturally wide. That measurement was not used to compute the cross sectional area.

The Atlas is 18 metres high. Was stated some time ago by the Dev team. Unless they changed that height, you will need to recalculate. Someone correct me if I am wrong. Edit: Thanks for the alternate data.

Edited by Dirk Le Daring, 14 January 2013 - 05:46 PM.