V-Ray Suggestion: VFB - Real Time, Camera, Lens and Film Attributes

[viscorbel]

+1 photoshop link would be useless for any 32 bit editing for the reasons mentioned.
I also agree about the way glare works in lensFX in vray - that ripple effect is unlike anything I've seen in years of doing photography. Not to mention the abrupt sharp cutoff of the rays when changing glare size, it just doesn't happen like that in real photos.

If you want to compete with streamlined post processing features of other renderers, then Photoshop live link would be very ineffective and unfortunate way of going about it.

First of all, it would not accommodate for vast majority of the things people want. It does not have any advanced glare solution, it doesn't have any CG friendly tone mapping, and most of all, it's extremely limited in terms of procedurality, let alone not being very fast in bitmap generating operations. So even if your render refined in realtime in Photoshop, it would likely not catch up very well. Also, anything Adobe-related is probably not worth investing into at the moment, considering it's starting to have some serious competition.

I would simply start with scraping down LensFX and doing a simple Corona/Octane/Fstorm style Glare. LensFX are based on some advanced technical paper, that does a lot more than a few primitive directional blur and color hue operations. In reward, LensFX is significantly slower, a lot harder to use, and most of all, fails at reproducing more than 95% of the glares generated by real cameras. It's a Loss/Loss/Loss/Loss scenario.

So I think it would be worth implementing a Glare generator that's a lot easier to implement, lot easier to use, lot faster, and generates glare effects that are lot closer to those captured by photo and video cameras.

About past year, I am noticing glares in all the TV series, Movies and Photography I watch. I don't think I've ever noticed a glare that had any prominent lens grating effect (that rippling LensFX has by default) and characteristics of the glares I observed were always something that could be achievable with Corona/Octane/Fstorm in just a few seconds with few clicks, or with LensFX after about half an hour of messing with trade-off between good looking glare and image bluriness, while also fiddling with very unintuitive external application, while still not getting it looking quite the way I want to.

[viscorbel]

I guess it comes down to this - there are 2 main types of users

A) People who want a 'photographic' image that comes out of the renderer and then do basic 8/16 bit postwork stuff on it.
By photographic, I mean image that looks like it was taken with a real camera - with the nice looking tonemapping (filmic wink wink), realistic bloom/glare, dof, etc already done.
Maybe like in actual camera you can do some preset color tweaks, use premade presets, LUT or whatever. Just add a bit of photoshop polish and color tweaks and send to client.

B) People who want 32 bit linear renders with all the passes and use them in advanced compositing software, doing tonemapping/glare/other 32 bit stuff themselves.

I think vray currently satisfies the B group, but the A group feels a bit neglected. Especially when the dreaded "C" word is doing all they can to cater to users like them.
And somehow I think there is a lot of folks in the A group, would be a shame to lose them to competition.

[dans]

My take is similar to Viscorbel, those effects that would greatly benefit from 32 bit space would be very helpful to have built in.

I just don't understand with a team and history such as Vray, Corona continually beats us to the punch with IPR, bloom and glare, Lightmix, etc? Typically there then is a discussion on the forums about a new thing, some people always says not in my personal workflow so we don't need it, Vlado feels it is not core render technology, and 6-12 months later after the topic refuses to die it gets implemented. Could we for once implement something user friendly first? Like the photoshop live link mentioned? PSD manager seems great but apparently does not support progressive renders, and no 2017 version yet so I can't test it, and this kind of compositing link with selection masks would really make sense to have integrated. I think the PSD developer will survive, he doesn't seem that active updating it.

psd-manager 4 updates are available - just not yet announced officially. Check http://www.cebas.com/forum/index.php?topic=566.0 (cebas login required).

psd-manager does support progressive renders, just V-Ray does not provide deep data to conserve memory. This affects EXR deep data just as well as far as I know. So you can save Render Elements, MultiMattes etc. as usual - just the mask generation tools in psd-manager require Adaptive. Some long time Beta users render their finals using Adaptive - not an issue then. If you miss way to stop a progressive render early, but still apply the render effects and save the image - that is an V-Ray issue. 3ds max 2017 solves this one because it has a STOP button in the Render Progress Dialog. Renderers that support it are supposed to apply render effects and save the image. It is just V-Ray not doing that correctly yet. I'm told it will be solved in the next service pack. However good news is, there is also a workaround in psd-manager 4 that will make this work right now also for older 3ds max versions.

Everyone would benefit from a standardized realtime effect Plug-in type in 3ds max. Every renderer implementing their same stuff just for their own framebuffer is not good for 3ds max. Typically if there was missing functionality in 3ds max it could be solved by plugins. Now that each renderer is cooking their own thing and trying to do the same thing for each 3d package in the same way - this possibility is falling apart. So the renderer developers battle this out among each other. I believe Autodesk is to blame for this development by not providing a better, extendable framebuffer when the time was right.

No standardized framebuffer -> no standardized UI -> no 3rd party plugins solving highly diverse needs.

Daniel

[Nicinus]

I guess it comes down to this - there are 2 main types of users

A) People who want a 'photographic' image that comes out of the renderer and then do basic 8/16 bit postwork stuff on it.
By photographic, I mean image that looks like it was taken with a real camera - with the nice looking tonemapping (filmic wink wink), realistic bloom/glare, dof, etc already done.
Maybe like in actual camera you can do some preset color tweaks, use premade presets, LUT or whatever. Just add a bit of photoshop polish and color tweaks and send to client.

B) People who want 32 bit linear renders with all the passes and use them in advanced compositing software, doing tonemapping/glare/other 32 bit stuff themselves.

I think vray currently satisfies the B group, but the A group feels a bit neglected. Especially when the dreaded "C" word is doing all they can to cater to users like them.
And somehow I think there is a lot of folks in the A group, would be a shame to lose them to competition.

This is probably a spot on description of the current market place. Although I would like to see myself as a B, my daily situation and everything else I have on my plate probably firmly places me in A. I wonder how this applies to everybody else on these forums and whether future growth for Chaos is in A or B, although I can't see why it couldn't be both with a clever interface.

[Alex_M]

I guess it comes down to this - there are 2 main types of users

A) People who want a 'photographic' image that comes out of the renderer and then do basic 8/16 bit postwork stuff on it.
By photographic, I mean image that looks like it was taken with a real camera - with the nice looking tonemapping (filmic wink wink), realistic bloom/glare, dof, etc already done.
Maybe like in actual camera you can do some preset color tweaks, use premade presets, LUT or whatever. Just add a bit of photoshop polish and color tweaks and send to client.

B) People who want 32 bit linear renders with all the passes and use them in advanced compositing software, doing tonemapping/glare/other 32 bit stuff themselves.

I think vray currently satisfies the B group, but the A group feels a bit neglected. Especially when the dreaded "C" word is doing all they can to cater to users like them.
And somehow I think there is a lot of folks in the A group, would be a shame to lose them to competition.

Wiser words have never been spoken. It does seem that Corona beats Vray in group A) (I still use Vray for 99% of the jobs tough). I myself fall in that group. I very rarely work with 32 bit outputs and fiddle with 32 bit compositing software like AE and Nuke. Most of the time I set up things to look good in the VFB and then just add some little finishing touches in PS then send to the client. That's my workflow in 99% of the time.

The main difficulties I'm facing with Vray at the moment are I guess these two:

1. Lack of a more photographic response to light similar to Fstorm and Corona. It feels that I have to struggle more and spend more time tweaking stuff to get a photographic feel from a Vray render compared to Fstorm and Corona. I'm not sure what "black magic" do Fstorm and Corona do but the interaction of light with the surfaces in these engines seem a bit more photographic, not that dry/CG'ish.

2. Strange behavior of Lens FX (ripples/banding patterns and sharp cut-off at the edges of the streaks). I recently posted a thread about this problem with screenshots and examples of how other render engines handle lens effects. I would also love it if Vray could generate the lens FX in real time during rendering like Fstorm and Corona.

[stevejjd]

I guess it comes down to this - there are 2 main types of users................

Thank you so much for these accurate, descriptive words. Clarification is everything.

I think the Photoshop link is a valid but separate issue of compositing.

Thank you as well to Nicinus and Alex_M, your comments were also very helpful in terms of clarification.

Vlado
We just want some nice Filmic response - I know there may be a lot of variety and its down to the look but just chose something and well tell you if its good. Just implement this one posted by viscorbel https://knarkowicz.wordpress.com/201...mapping-curve/

[vlado]

I just don't understand with a team and history such as Vray, Corona continually beats us to the punch with IPR, bloom and glare, Lightmix, etc? Typically there then is a discussion on the forums about a new thing, some people always says not in my personal workflow so we don't need it, Vlado feels it is not core render technology, and 6-12 months later after the topic refuses to die it gets implemented. Could we for once implement something user friendly first?

This is such an inaccurate statement to make. You got things like the distance texture, triplanar texture, clipper, hair, stereo+VR rendering and many others way before they got into Corona. Some of them even happened without any forum discussions at all It is normal that different render engines will have different feature sets simply because the developers have different priorities.

The other reason why I delay these things is a little more sneaky, and it's because the other renderers don't always get things right. I like to wait a little bit and see what kind of issues and problems will come up for them before spending time and effort on a big feature. For example, you may or may not know, but Ondra had some questions on the 3ds Max beta board about how to implement IPR and I made some suggestions. I was not sure if they worked or not, so I wanted to wait and see how things will go for him. Once I knew that it worked fine, I could be sure that I could use the same approach for V-Ray's IPR. 6 to 12 months seems to be the right amount of time to wait. After all, these guys have looked at all the mistakes we've made so far and acted accordingly. Why shouldn't we be allowed to do the same?

Best regards,
Vlado

[LudvikKoutny]

This is such an inaccurate statement to make. You got things like the distance texture, triplanar texture, clipper, hair, stereo+VR rendering and many others way before they got into Corona. Some of them even happened without any forum discussions at all It is normal that different render engines will have different feature sets simply because the developers have different priorities.

The other reason why I delay these things is a little more sneaky, and it's because the other renderers don't always get things right. I like to wait a little bit and see what kind of issues and problems will come up for them before spending time and effort on a big feature. For example, you may or may not know, but Ondra had some questions on the 3ds Max beta board about how to implement IPR and I made some suggestions. I was not sure if they worked or not, so I wanted to wait and see how things will go for him. Once I knew that it worked fine, I could be sure that I could use the same approach for V-Ray's IPR. 6 to 12 months seems to be the right amount of time to wait. After all, these guys have looked at all the mistakes we've made so far and acted accordingly. Why shouldn't we be allowed to do the same?

Best regards,
Vlado

It's incredible how despite the forum being filled with Corona this and that things recently, you still manage to stay cool, fair and positive That really tells a lot of (in this case very good) things about one's personality

[Alan Iglesias]

It's incredible how despite the forum being filled with Corona this and that things recently, you still manage to stay cool, fair and positive That really tells a lot of (in this case very good) things about one's personality

So true!

Working closely with GaryY and his team during the formative years of 3D Studio DOS and 3DS Max taught me that Really Good Software is really more about the folks who code and develop it than anything else. Dedication, competancy, and plain old hard work trumps a feature here and there every time and heck, we know we'll get it most likely sooner than later.

And this why sticking with Vray makes the most sense to me and why I highly recommend it to all my clients.

Just my .02$...

-Alan

[Nicinus]

This is such an inaccurate statement to make. You got things like the distance texture, triplanar texture, clipper, hair, stereo+VR rendering and many others way before they got into Corona. Some of them even happened without any forum discussions at all It is normal that different render engines will have different feature sets simply because the developers have different priorities.

The other reason why I delay these things is a little more sneaky, and it's because the other renderers don't always get things right. I like to wait a little bit and see what kind of issues and problems will come up for them before spending time and effort on a big feature. For example, you may or may not know, but Ondra had some questions on the 3ds Max beta board about how to implement IPR and I made some suggestions. I was not sure if they worked or not, so I wanted to wait and see how things will go for him. Once I knew that it worked fine, I could be sure that I could use the same approach for V-Ray's IPR. 6 to 12 months seems to be the right amount of time to wait. After all, these guys have looked at all the mistakes we've made so far and acted accordingly. Why shouldn't we be allowed to do the same?

Best regards,
Vlado

Lol. My apologies, I was probably in a bad mood when I wrote that. Definitely not fair. However, I seem to be starting to realize that I am a type A user and it seems that you as a developer, with different priorities, have a focus on the B category. Now, I'm not interested in Corona as I think the future belongs to GPU solutions, but they do seem to focus more on the category A features, which makes me a bit envious. Of the features you mention above I've only used the stereo+VR rendering (that I'm aware of .

[vlado]

Now, I'm not interested in Corona as I think the future belongs to GPU solutions, but they do seem to focus more on the category A features, which makes me a bit envious.

I understand that; hopefully we can make V-Ray match that experience in one way or another.

Best regards,
Vlado

[vlado]

Something else that I forgot to mention above... it is difficult for me to distinguish the hype surrounding a given feature vs its actual usability in practice. So, it makes sense to wait a certain amount of time to see if something would be really useful or just a momentary blip. For example, a couple of years ago, we had lots of people asking about OSL and shading languages in general. It's a huge feature that took us a lot of development resources. But in the end, turned to be used only by a handful of people. So lens effects are all the rage right now, that's fine and obviously something that we should take into account, but whether a few months from now that would still be the case... I guess we'll see once we do it.

Best regards,
Vlado

[LudvikKoutny]

I guess in case of Glares it won't go away. It will become standard I think. People strive for photorealism and it is impossible to capture a picture with camera that will give you sharp, burned out aliased highlight without any glare or bloom like optical effect. It's just another step to ultimate photorealism I think, in a same way for example caustics.

People use them rather rarely now, but if caustics were automatic, without any need for user intervention by setting up photon mapping and special emitters for caustics so that they converge reasonably well, people would expect them on every image.

So now that glares are something that's a no-brainer to set up, I see them becoming a standard, rather than a gimmick. It's also a missing piece of puzzle that allows people to finally let go of things like subpixel mapping. Hotspots always looked unnatural as they were missing those secondary effects.

EDIT: Also, the fact that developers of certain unsuccessful CPU and then GPU renderer managed to make a business out of selling Photoshop plugin that does just glares, CG friendly tone mapping and denoising/firefly removal, and that this business is thriving, suggest that this may be more than just a temporary hype

[vlado]

I guess you are right...

Best regards,
Vlado

[raymarcher]

I agree with Recon, Photoshop is a pita to do lens effects.