Page 1 of 1

ideas for a more powerful implementation of HDR-Light-Studio

Posted: Sat Mar 19, 2011 10:30 pm
by Polyxo
Hi Developers,
I find HDR-Light-Studio a very interesting Software. After a long discussion in this Thread I also understood that its way to deal with Scene-Lighting
is the only one which currently allows for interactive Light-Source relocation without need for Maxwell to re-voxel the Scene.
Both Plugin-native Lights as well as Emitters applied to meshes currently require revoxeling as soon as we touch them which may decrease the Joy
of interactive Rendering with Fire considerably - at least with heavy Scenes.

Of course I have no idea at all if this Limitation is there to stay - but for now HDR-Light Studio seems to be the only option to really have
interactive influence on Light positioning resulting in immediate feedback in the Scene. Also the program ships with a lot of interesting
HDR-Light-Sources. The only thing I always found awkward was its GUI. While it now got greatly more accessible with the direct
implementation to M~Studio it still forces us to deal in 2D with an unwrapped Environment-Sphere. We have to push Lights around
on a black Canvas without actually seeing our already perfectly set up 3D-Scene. I found this neither intuitive nor fitting well the
Maxwell-Paradigm which always tried to give use a powerful translation of familar aspects of physical cameras.

Attached are some Sketches which illustrate how I could imagine to make this Plugin to Maxwell more useful and feeling more integrated.
I basically assume that one can convert Lights to Cameras temporarily how this is already possible in some supported 3D-Apps
(3DS-Max/Rhino-probably others too). While in Light-View one can aim/reposition the Light-Source with the normal Camera-Controls,
one can configure Falloff and in the case of making this useful for HDR-LS one also had to have an image-loader to exchange the
Light-emitting texture. I imagined the Hdr positioned interactively in the Scene by the user getting projected accordingly
to the same virtual Environment-Sphere visible when currently using the HDR-LS-Plugin.

Opinions?

Holger

Image

Image

Re: ideas for a more powerful implementation of HDR-Light-St

Posted: Mon Mar 21, 2011 11:18 am
by Polyxo
No takers?
I'd at least love to hear, whether such an implementation was theoretically possible, regardless of effort
of bringing something similar to what I proposed to Studio or to Plugins with a unified GUI...

I mean - Fire is a super-fantastic addition and I am very happy about it...
But one at this time actually needs a very expensive Plugin with questionable implementation to make any interactive
changes on Light (appart from rotating content-wise static HDR-Backdrops).

I'd basically like to know whether this is basically the way it's supposed to stay: So that the proposed way for anyone
who wants more interactivity in Light-Setup is purchasing a copy of HDR-Light-Studio?
Or whether it is the plan for upcoming releases to hook up Mesh-Based Lights without need to Revoxel the Scene?

So that any Maxwell user can relatively soon profit from the same kind of Setup-Interactivity as offered by the 3rd party Plugin
when using conventional mesh-based Lighting (minus the cool Image-Content offered by the Plugin obviously).

Re: ideas for a more powerful implementation of HDR-Light-St

Posted: Mon Apr 04, 2011 1:14 pm
by dariolanza
Hello Polyxo,

Interesting.

Option 1: changing the voxelization rutines to avoid different voxelizations when moving emitters is simply not possible. Any moving object may need a new voxelization to inform the system about the new position of the objects.

Option 2: we studied the possibility of developing our own HDRI editor inside Studio, but as Studio has too few users, and we are involved in new complex developments of the core engine, it is not possible for us to crete it now, and it is not in our current roadmap.

Cheers

Dario Lanza

Re: ideas for a more powerful implementation of HDR-Light-St

Posted: Mon Apr 04, 2011 1:31 pm
by Polyxo
Thank you Dario,
I'm glad for an official reply!
Option 2: we studied the possibility of developing our own HDRI editor inside Studio, but as Studio has too few users, and we are involved in new complex developments of the core engine, it is not possible for us to crete it now, and it is not in our current roadmap.
Understood. What I was thinking of actually was adding one Maxwell -Light-Type which was made available both in Studio and inside the Plugins
possibly by modifying an existing Cg-Light-Type and reuising its placement-widgets.
It should be possible to place it in the way one would place a CG-Light but also allow for this Light-View I described.

This proposed Maxwell-Light however would not act as a typical Mesh-Based Light-Source but in fact use the same tricks which allow
HDR-LS for editing the Environment-Sphere in Realtime. Have a look at this and imagine that the mesh from the Lightboxes
was entirely ignored by the Maxwell-Engine. It was however realtime-projected to the Environment-Sphere and evaluated as preliminary maps
hooked up in the Environment-Slots. No idea what Voodoo HDR-LS does here exactly but it seems to work. I'd just wish for a proper interactive
3D-Implementation done by you Guys,especially as the Revoxelization-Issue is there to stay.

Re: ideas for a more powerful implementation of HDR-Light-St

Posted: Mon May 09, 2011 7:00 pm
by Seggy
Hi Holger (Polyxo),

I am from the Lightmap team who develop HDR Light Studio.

You have some interesting ideas, but I think you are ignoring what is the main benefit of using HDR Light Studio.
The HDR canvas interface allows you to place a light anywhere in relation to your subject matter - but driven by a single 2D view.

Once you have your camera angle set up and the FIRE view working, you can light using our 2D interface very easily and very fast and concentrate simply on the content of your render. No messing around in 3D moving emitters etc.

Now the main problem you outline is understanding where to place the lights on the HDR canvas, and you want to go back to the traditional method of working in 3D around your object and seeing them around your object. But placing lights is not a random process in HDR Light Studio, once you have a little experience then it is very easy to know where is to the left, right, top and bottom of your object. From here it is simple to place and adjust lights where you want them. Plus the feedback is instant.

When using HDR Light Studio in many ways you do not care 'where' the lights really are, but what your object looks like in the render view. To be honest when working in true 3D and placing your emitter lights, it is still a guessing game where to place a light in 3D space in order to get it reflected in the part of your model where you want it to be seen. Saying that, if you are well practiced at anything, it seems easy to you.

So I feel if you used the software for some time, you would not feel the need to try and mould it back into a traditional solution, but you could embrace this new way of working.

I think it is unfortunate that because the HDRI updates in real-time in FIRE and the emitters do not at present, then somehow the HDR Light Studio solution is being looked at as a real-time alternative to emitters, used to solve this problem. Our software is much more than moving around solid lumps of geometry (like Softboxes). With HDR Light Studio you can enhance existing HDRI backgrounds with soft lighting effects and overlap lights for soft reflected lighting effects. It’s an advanced image based lighting tool, and image based lighting can do so much more than mimic geometry lights, that you can easily build and move around in 3D for yourself. Using IBL also happens to render pretty fast, which is useful.

I like your ideas for deeper integration and understand why you want them, but the thin and simple integration at present works very well - if you can just get over the hurdle of learning this new approach. From our interface you can see we like to make things simple.
I hope this alternative point of view is helpful.

I hope that emitters one day update in real-time in Fire so we can move on to look at the real benefits of using HDR Light Studio. I think that real-time emitters is really the solution you are looking for.

Best Regards

Mark

Re: ideas for a more powerful implementation of HDR-Light-St

Posted: Mon May 09, 2011 10:33 pm
by Polyxo
Hi Mark,
thank you for taking the time!
Trust me, I've tested your Software intensively but I honestly can't agree with your perception.
I wrote about my reasons in great length in this thread too.

May I ask back - have you already worked with a Software which allows for the sort of Light-View I describe? (e.g. 3DSMax)
What you call "messing around with Emitters" is what I call the most ideal concept of Setting up Scene-Lighting.
Forget dragging Lights into place with Translation-Widgets - just set them up as if you were looking through the
Light itself. See the results imediately in the Viewport or Fire. I prefered this Approach over HDR-LS without having
to think for a mere second. If that was different I certainly hadn't sat down to draw these Graphics.

I'm not out to bash your Software, but I'm honestly not convinced. The 2D-Concept of its current Incarnation, the Fact that
one has to use Studio but then again leave it for a non integrated Plugin as well as its overall User-Experience to me seem not elegant
enough for the Host-Application.
What to me seems most disappointing is the fact your Software boils Lighting down to a Bitmap driven affair.
That may well be so also under the Hood also when using CG-Lights Emitter Gobos and all that stuff. But it doesn't feel so damn prosaic.
Maxwell does the best to mimic a physical Camera - you deliver a Light-emitting flat Photo-Wallpaper which - even worse - also feels like a
Photo Wallpaper when setting it up. Having your Relighting Technology available inside Emitters which have Location in Space would make
a World of a Difference.

Its cool though that you make a tool available which at this point is the only option to circumvent a huge shortcoming of Fire when
compared to other Interactive Renderers in the Market: Missing interactive Relighting.
I'm sure for that reason alone HDR LS has already proven valuable quite a few Maxwell-Customers.

I think however having to use an expensive 3rd Party Software should not stay the only option to get what Products competing with
Maxwell deliver as inbuilt Feature. Actually I am surprised that People stay that calm...
Next Limit really should close this gap and give users who need true interactivety the Choice which method to take.
But that's where we two agree :)

Greetings, Holger

Re: ideas for a more powerful implementation of HDR-Light-St

Posted: Tue May 10, 2011 9:12 am
by Seggy
Hi Holger,

You don't want to light your scenes using an interactive HDRI or using HDRI at all.
That is what out software provides - its a 'HDR Light Studio'. It is really good at this, that's what it is designed for.
Look at your FIRE window and what you see there is what you get. Dont worry about the HDR canvas.

You want emitters in 3D space working in realtime.
Stating this opinion is not bashing our software.

I think a better title for your posting woud have been.
'Ideas for a studio lighting system in Maxwell Studio.'

You want an alternative Holger, not to change our solution.
You want something like SLIK in Modo, but realtime in Maxwell.

We have been in business for over 2 years developing and selling this lighting solution before we got compatibility with Maxwell. We have become the industry standard as a HDR lighting design tool. Some of our customers asked us to try to make this compatibility happen with Maxwell. You can see from our web site (www.hdrlightstudio.com) that lots of users enjoy making pictures with our software and that the results are beautiful. We are very proud of our gallery with renders made using a wide variety of renderers. Users report being able to light their scenes much much faster than before, and with faster final render times using HDRI. For many our lighting tool is now at the heart of their everyday lighting workflow. If the solution was bad, I do not feel Next Limit would want to provide compatibility with a bad solution - we do exactly what it says on the tin.

Re: ideas for a more powerful implementation of HDR-Light-St

Posted: Tue May 10, 2011 12:21 pm
by Polyxo
Hi Mark,
You don't want to light your scenes using an interactive HDRI or using HDRI at all.
Sorry but this assumtion is not correct. As probably all Maxwell Customers I see the value of HDR-Lighting
and use it very often. As emitting Environment as well as mapped Emitter.
Indeed I suggested an alternative GUI for your Software which I consider superior and better integrateable
also to Plugins. What I want is an option to orient Light-Sources by temporarilly turning them into something
close to a Camera, let this be any HDR-Lightsource (or at some point) a classic unmapped Emitter.
I want to see the Position on Space as well as the Contribution to Scene-Lighting in Realtime.

You may of course not like the idea.
You want an alternative Holger, not to change our solution.
You want something like SLIK in Modo, but realtime in Maxwell.
Again oh please no.
Actually having detailed Tripods and Rails/Fixtures for Studio Lights is something I consider horrible Kitsch in a
3D-Environment. What I'm interested in in maximized Performance and not Man's Toys in Dollhouse-Format.

Holger

Re: ideas for a more powerful implementation of HDR-Light-St

Posted: Thu May 12, 2011 2:48 pm
by Seggy
I understand you better now. Which is good news.

I did an article on size and closeness in a HDRI, http://www.hdrlightstudio.com/blog/hdri ... /#more-237

I think knowing the direction of where the light is may be important and seeing from the lights view is kind of cool - but how close it is to the object is not important for a HDRI. A light can be small but very close, or large but far away and appear just the same in the HDRI view.

We will think about your ideas.
Nice to talk to you.

Re: ideas for a more powerful implementation of HDR-Light-St

Posted: Fri May 13, 2011 3:03 pm
by Polyxo
Good to see that we both got our points across Mark!

What I believe is always a challenge in creating GUI's for Computer-Graphics is:
How close to comparable Real-World Activities should Touch and Feel get?

As already discussed Slik would be an example which totally doesn't work for me:
Lights on Tripods (as Tools, not as Scene-Members) in an Environment which doesn't know Gravity?
Such just appears weird to me.
Other examples could be typical Audio-Editing Software. They come with imitations of shiny turned
Aluminum-Knobs and heavy Shadows are supposed to make Sliders look 3D. While I find such GUI's plain
horrible to look at I can not deny that the 1:1 mapping between Software and Physical Mixers
clearly helps - anyone who can use the Hardware-Version is up and running with the digital Equivalent in no time.

Maxwell in my perception choses a delicate Balance: There's proven aspects of Physical Cameras every
Photographer will recognize and not have to relearn fused with a very powerful and straightforward to
use excerpt of all known strategies of rendering images digitally.

This Setup provides me as a User with the Feeling to have a very powerful Studio-Camera which (in contrast
to many other Renderers) doesn't require me to be an Engineer to use it but lets me concentrate on making a Picture.

While digital the whole process feels very craftsmanly and at every stage 3-dimensional.
Ok, it gets represented by a 2D Screen but I guess we all feel that we are working inside a Room and not on a Canvas.

When I have to create and adapt textures for my models I for the same reasons use 3D-Painting Apps
and draw on the Target-Object directly. Also here I tend to avoid 2D-Tools like Photoshop as they hide one
highly important dimension from my Eyes. While 3D-Painting-Apps of course also write 2D-Textures I simply consider
them the superior concept for setting Material up for Render-Work.

Now with these thoughts as a background I believe you understand why I dislike the 2D-Canvas Idea of your Software,
although I clearly understand that in the End it does not matter how far or distant a Light-Source is from an Object when
it gets rendered into a Hdri.

I could always find out quickly and instinctively (=by using existing experience) where one needs to place a Light
in a physical Room to achieve a certain Effect but even the World best Photographer had no Clue where to position that Light
when asked to draw its position and size it onto a piece of Paper. The Feedback your Software gives helps greatly but that
doesn't cure the fact that we all start "dumb" when using HDR-LS.

At that point in my opinion proven existing User-Knowledge gets disregarded.
We instead have to learn a new, without any doubt more abstract representation instead of getting a super easy to handle interface
to harvest our Experiences in positioning Lights. This although our Struggling is effectively not caused by not knowing the optimal
Light-Location but by clumsy 3D-Manipulators and nonexistent realtime Feedback of Lighting with default Maxwell Lightsources.

My idea should basically use the principles of your Software but transferred to a GUI which I believe is closer to
a Photographer/Image-Composer thinks.

cheers, Holger