Modding Summary Thread : What we Know so far!

Posted on the original forums. Thought you guys would be interested in this. Note they are still predictions.

Having digested the recent information I think its only fair to make a summary thread of information regarding modding. Please can this thread be updated and not used as a place to debate but rather collect useful information and theories realistic possibilities.

Modeling

3d Studio Max is the main 3d app - Little slip by Helios with the new PLA uniform render. What does this mean? That either a plugin will be provided or need to be scripted for 3ds Max.

Polygon counts - Judging by the recent renders and past renders budgets seem to follow as such (Of course this is speculation but is based off several professional opinions)

Weapons (No attachments) - 1000-8000
Character (No Heads or Body parts) 5000-12000
Attachments - 1000-5000
Vehicles - 15000 - 30000
Buildings - 5000 - 20000

Take these with a pinch of salt and as a very rough guide line, the engine probably can support much much higher (cars in GRID for example at LOD1 are 20-40,000 tri's)

Vehicle Models Appear to be make up of several individual parts or segments - This would suggest to me that the models are connected in Max via proxy/nodes or through a 3d party app (possibly CM's own tools) Advice to modelers is to section up models after being welded.

Character Modeling Is possible similar approach is taken when modeling humans with regards to wounds.

Models are interchangeable Judging by the blooper in the Ego tech demo and the previously announced feature of customizable weapons it would appear that segments of the model are called separately and are unlikely to be whole models. Eg A base M16A4 would be used with the different attachments being different models either called via a script ingame, or contained hidden within the base models animation or/and as a proxy (like Crysis)

This might also apply to bodies and in particular equipment carried on bodies.

Animation Its possible that canned animation is purely object based and only the object needs to be animated opposed to all animations be saved within a scene (like Crysis). There is likely a Base skeleton with animations exported out, saved and called ingame via a script. This probably would make most sense giving the number of units and may work in a similar fashion to HL2 though much more flexible.

Facial animation Thinking again, give the use of the new technology facial animation would be controlled with a set rig, emotions set and exported out and called in when needed via a script rather then complex scenes being used. Am I going out on a limb here but I'd hazard the guess they maybe a 3rd party tool for Max or even a full program (Like Face Generator or HL Poser) used.

Textures

DDS is the likely extracted format - This is based off research from GRID/DiRT formats and the fact DDS is the most common way of building a compressed Normal/Spec map.

Dimensions I'd hazard the guess the upper limit at being 4096* though for good performance most likely dimensions will be 2048* or below.

Scripting, Coding & File Format

LUA is the scripting language! Learn it!

XML / Encoded Script might be used This is based off research of GRID though it may not be the case with OPF2 its certainly used in GRID (An EGO powered game).

Node based scripting In the EGO tech demo its briefly mentioned when a dev changes the smoke colour, though its not actually seen or mentioned any further - In-Editor scripting make work in a similar way to Crysis Sandbox node scripting.

PSSG File format Used by both GRID and DiRT, it houses the model and texture files of content, maybe be the same format used in OPF2.

Script based Animation Animation in the non-phyiscs sense might be script driven, ie there is a base animation which is applied to objects when needed.

Tools

There is an official Mission Editor How much it can do is unknown however it appears pretty extensive with options to create custom groups and test real time. This is speculation but by the definition of different fireteam types it appears to me that there is either;

Ai specific to certain roles/groups eg AA soldier
There is kit or load outs controlled via a script (This may also control interchangeable model segments)
Behavior is governed by the type of group

Streaming Tech This would indicate that they are multiple levels of data for terrain to create procedural (sp) terrain on the fly (Similar to how Pure works) Thus I can only guess that in the creation of a new terrain only the height data would be needed as the engine would take care of the rest.

LODs Maybe be totally procedural however if this is not the case then multiple LOD models will exist for every object.

Damage Model Similar theory however they may exist several variations of the same damage model though called in randomly for each state of destruction.

Physics at this stage its unknown how physics can be applied though there are a number of possible solutions I personally think it boil down two ways of implementation.

A physics library where physics are defined in Code and in turn called via script to control the behavior of an object (GTA IV and HL2 use this approach)

or

Physics applied within application (Probably Max, similar in respect to Crysis or UT3) with physics acting and reacting to how there configured in 3d application or third party tool.

Sound Totally unknown at this stage but 99% of games use OGG, WAV and other common uncompressed file formats before compressing them into the property file format (Can someone verify this with GRID/DiRT?)

Ai Again totally unknown, most likely the 'playbook' is hardcoded with behavior defined via Nodes or LUA scripting in the Editor / Game Environment.

Cross Platform, DLC and Mod management

No information what so ever on these aspects however given that is a cross platform title (in particular the Xbox360) I'd imagine the folder structure to be very tight and would be pleasantly surprised if Mods/Addons can be called without the need to over write, at least in the early stages of modding. One of the first tasks if this isnt the case would be to look to find ways to create substructures, LUA is very powerful and may well let us do this.

Does it take a long time?