Software Engineer needing some Architectural Insight

[matthewnorman]

Hoping this is the correct area to post.

 

Hi,

 

I'm a software engineer working on a program to help architects take a base model and generate a real time(using games tech) and more traditional rendered output(video, stills etc). This includes adding materials/lights etc.

 

I was hoping that you guys could help me better understand the preferred work flow for this(we were recently told by a revit user that our material process felt backwards) and what kind of outputs are preferred, i.e is oculus rift useful, 8k stills or 3d movie/walkthroughs to name a few.

 

I come from a more games orientated background so any insight into how architects work would be greatly appreciated.

 

If anything I've said doesn't make sense/isn't clear just let me know and I'll elaborate.

 

Again any thought people have to offer will be really useful for me.

Thanks.

[numerobis]

Looking at something like Lumion3D you could maybe focus on more realistic materials and lighting...

Brigade is something that is heading in this direction but using path tracing http://brigade3.com/brigade-3-0-preview-at-gtc-2014/

[beestee]

Maybe you could get some general ideas from an interface that is already working for architects by taking a look at some software demos of existing products that architects use. I am not going to say you should rip somebody else's work off, but it could make your path a little more defined and give you some ideas for where to start...after all, you have come to a community that continuously improves their own output by learning from the work of others.

 

Honestly, Revit's material interface is a little bit backwards on it's own, far from what I prefer. It seems most software that you would be competing with have exporters for Revit that will "read in" the materials and lights along with the model info. Not sure why a user would comment that your interface is backwards since going realtime with a Revit model is typically known and understood to be a one-way street.

 

How are you handling materials now?

[matthewnorman]

Looking at something like Lumion3D you could maybe focus on more realistic materials and lighting...

Brigade is something that is heading in this direction but using path tracing http://brigade3.com/brigade-3-0-preview-at-gtc-2014/

 

Yeah, our underlying engine is Unity3D, so we're very excited about being able to implement some of the new features in 5.0, hopefully things like the PBS should make materials a lot easier as its likely it will be the only shader we need. Lighting at the moment needs a bit more work we're looking at hopefully being able to support ies lights.

 

Path tracing is really cool, our biggest issue with it would simply be the time it would take to implement, buts its definitelysomething we should look at.

Cheers.

[matthewnorman]

Maybe you could get some general ideas from an interface that is already working for architects by taking a look at some software demos of existing products that architects use. I am not going to say you should rip somebody else's work off, but it could make your path a little more defined and give you some ideas for where to start...after all, you have come to a community that continuously improves their own output by learning from the work of others.

 

Honestly, Revit's material interface is a little bit backwards on it's own, far from what I prefer. It seems most software that you would be competing with have exporters for Revit that will "read in" the materials and lights along with the model info. Not sure why a user would comment that your interface is backwards since going realtime with a Revit model is typically known and understood to be a one-way street.

 

How are you handling materials now?

 

At the moment when importing we use the name of the material to form what we call material groups. These work by assigning objects to groups and every object in the group will use the same material. I think where we lose/confuse people is that what you actually edit is a material in a group and not a material on an object (though your changes update live on your model). It is possible to create new groups and merge existing groups, however I don't think that we have done a very good job of making this obvious to the user.

[M V]

If it is going to be geared toward architects, it should be able to wait last minute to make changes, be indecisive, create crappy 3D models, and have a button to create 'MORE POP'.

 

All joking aside, it should not be overly complicated with render settings like for example, "refraction" and "fresnel". Architects don't know or care about this. I think a hybrid of Lumion, Colimo Motiva, and Cl3ver would be the goal. It needs to be

1. EASY and INTUITIVE

2. CHEAPER than Lumion

3. INTEGRATE with architects models from SketchUp, Revit, etc. EASILY

[Devin Johnston]

Looking at something like Lumion3D you could maybe focus on more realistic materials and lighting...

Brigade is something that is heading in this direction but using path tracing http://brigade3.com/brigade-3-0-preview-at-gtc-2014/

 

This is the first I've heard of Brigade and while I like the visuals it looks to me to be on the same level as Vray's or Maxwell's real time rendering solutions. Unfortunately noise is the big problem here so unless it can be cleared quickly it's only adding to the already large pool of path tracing options out there. This is the problem I have with real time walkthrough solutions, either they are to cartoonish like Lumion or Twinmotion or they have too much noise which is distracting.

 

Matthew my first question is who are you creating this program for, Architects or architectural illustrators?

[matthewnorman]

If it is going to be geared toward architects, it should be able to wait last minute to make changes, be indecisive, create crappy 3D models, and have a button to create 'MORE POP'.

 

All joking aside, it should not be overly complicated with render settings like for example, "refraction" and "fresnel". Architects don't know or care about this. I think a hybrid of Lumion, Colimo Motiva, and Cl3ver would be the goal. It needs to be

1. EASY and INTUITIVE

2. CHEAPER than Lumion

3. INTEGRATE with architects models from SketchUp, Revit, etc. EASILY

 

1. This is something we're working on quite a bit at the moment, I just today have been adding in analytic s so we can get a better understanding of how people use the software.

2. Well we got that one covered. YAY!

3. This is actually been a real hot topic in the office recently, at the moment we import an fbx, but if half way in our process you need to make a change back in your modeling software it would mean you'd need to start again in ours. We have some editing tools but obviously we never are nor want to have the same amount as something like max.

[Scott Schroeder]

3. This is actually been a real hot topic in the office recently, at the moment we import an fbx, but if half way in our process you need to make a change back in your modeling software it would mean you'd need to start again in ours. We have some editing tools but obviously we never are nor want to have the same amount as something like max.

 

#3 is 100,000% critical you make sure you have a near seamless integration. Architects can and usually do change things at the last possible second. Designs are never finalized until the deadline has passed. If you have to restart all over again, your product is going to be a failure (if you are going to specifically target architects) before you even sell one copy.

 

Frankly, the market is already overloaded with sub-par, plastic looking real time arch viz "easy button" solutions that cost a stupid amount of money for what they can actually produce. Saying you are cheaper than Luminon is at the moment, since we don't know your price, is like saying that Bentley you are trying to sell is at least cheaper than a space shuttle.

 

For me, unless you can beat UE4's $19 a month price with no royalty fees for architecture products, I might as well stick with Autodesk Showcase that comes with our subscription package. It gives me the middle to lower end of the real time quality look and it's completely integrated into Autodesk products.

 

If I am going to leap into real time, I'll want to do it right with a higher end engine like UE4 that has physically based materials and lighting. I need something that is going to be as close to a stock Vray render as possible, because I can't go show my client a nice clean rendered image and then show them the same project done is some clunky real time engine.

[Devin Johnston]

If I am going to leap into real time, I'll want to do it right with a higher end engine like UE4 that has physically based materials and lighting. I need something that is going to be as close to a stock Vray render as possible, because I can't go show my client a nice clean rendered image and then show them the same project done is some clunky real time engine.

 

+1

[matthewnorman]

#3 is 100,000% critical you make sure you have a near seamless integration. Architects can and usually do change things at the last possible second. Designs are never finalized until the deadline has passed. If you have to restart all over again, your product is going to be a failure (if you are going to specifically target architects) before you even sell one copy.

 

Frankly, the market is already overloaded with sub-par, plastic looking real time arch viz "easy button" solutions that cost a stupid amount of money for what they can actually produce. Saying you are cheaper than Luminon is at the moment, since we don't know your price, is like saying that Bentley you are trying to sell is at least cheaper than a space shuttle.

 

For me, unless you can beat UE4's $19 a month price with no royalty fees for architecture products, I might as well stick with Autodesk Showcase that comes with our subscription package. It gives me the middle to lower end of the real time quality look and it's completely integrated into Autodesk products.

 

If I am going to leap into real time, I'll want to do it right with a higher end engine like UE4 that has physically based materials and lighting. I need something that is going to be as close to a stock Vray render as possible, because I can't go show my client a nice clean rendered image and then show them the same project done is some clunky real time engine.

 

At the moment we offer the ability to translate, rotate, scale, delete and duplicate Objects. We also include material and light editing as well as the ability to add some generic objects into the scene (trees, chairs etc) and of course you could import any furniture type models you already have. We included these to try and resolve the issue of late changes, but do you think these are kinda useless features if they don't get applied to the original model (fbx you imported)? We have spoken about basic mesh editing would this appeal or is it more a case of I'd rather just use my modelling software for that?

 

In terms of the market I think your spot on, I'm not sure how some people can live with below 5fps to do anything. At the moment we're pricing at £299.00 one of purchase or £27.00 monthly with access to our live/cloud services, both of these are royalty free. you can check out REALIS3D for context on the price. At the moment were still not fully featured, so I think the price is more aimed at "potential product", but do you think that we're pricing ourselves out of the market?

 

In terms of visuals I also think your spot on and we're currently looking at a number of lighting solutions (we use unity) and have yet to include PBS but have them ready to implement, but we're just burning down the backlog at the moment.

 

BTW thanks for the feedback this is really helpful.

[harryhirsch]

If it is going to be geared toward architects, it should be able to wait last minute to make changes, be indecisive, create crappy 3D models, and have a button to create 'MORE POP'

...you seem to forget that this is webpage is for professionals and that a lot of the members are architects. Please save us the childish gossip and have a little bit more respect for your clients.

[matthewnorman]

This is the first I've heard of Brigade and while I like the visuals it looks to me to be on the same level as Vray's or Maxwell's real time rendering solutions. Unfortunately noise is the big problem here so unless it can be cleared quickly it's only adding to the already large pool of path tracing options out there. This is the problem I have with real time walkthrough solutions, either they are to cartoonish like Lumion or Twinmotion or they have too much noise which is distracting.

 

Matthew my first question is who are you creating this program for, Architects or architectural illustrators?

 

Both, though its aimed at users with less technical knowledge, especially when it comes to realtime demo's alot of people don't want to learn(at least we hope they don't) how to use engines like Unity3D or UE. We can offer a range of different styles on the same model through post-processing. It's also a lot less complicated then something like max.

[Scott Schroeder]

I think the best world would be as Benjamin mentioned, the ability to read in materials from the FBX or whatever file. I'd really like to see it work like the Revit link inside max where if I refresh the Revit link, I get the updated model but all of my custom Vray materials stay on the objects they are applied to.

 

I'm all for making an easier way into real time, but if you try to dumb it down too much so that John Q. Renderman can use it, then you are going to get dumb results. Your target users should be people who at least give 1/3 of a crap about learning the process of real time rendering.

[numerobis]

Yeah, our underlying engine is Unity3D, so we're very excited about being able to implement some of the new features in 5.0, hopefully things like the PBS should make materials a lot easier as its likely it will be the only shader we need. Lighting at the moment needs a bit more work we're looking at hopefully being able to support ies lights.

 

Path tracing is really cool, our biggest issue with it would simply be the time it would take to implement, buts its definitelysomething we should look at.

Cheers.

 

Sounds good! I'm looking forward to see some results...

 

 

 

And yes, file linking or an easy and robust way to update geometry is important.

Edited May 24, 2014 by numerobis