pouët.net

hello everybody,

„Why do you torture yourself with such a difficult medium? Why not just use modern offline renderer? „

from my (a 3d artist guy´s) point of view -

the boundaries between real-time and "normal" rendering disappear - or will soon disappear completely.

and real time makes things easier to set up. eg. lighting situations instead of the need for waiting for it
to render 1 hour+. it gives you freedom design wise. how does this look with eg. a volumetric light ?

the concepts and principles and the knowledge you need for doing this - animation, knowledge about staging, modeling / sculpting, blend shapes setup, animation, anatomy, cinematography, lighting, texturing - related knowledge about uv coordinates, rigging, compositing etc. etc. etc. - all non-coder stuff.

long story short - it´s the same amount of work in both worlds if you want to do it right - realtime or not.
rt has a lot of benefits - so people choose rt.

don't tell me that this is less effort to learn or that it's worth less than doing „coding kung fu“.

I compare this to the film „some kind of monster“ by Metallica:
the dispute between lars ulrich and hetfield:
"fuck - all these fucking rules - this is a fucking metal band !!“

just like here: what are all these rules about? the demoscene has its roots in the cracker scene - so please people

these are - from my point of view - from the point of view of a 3d artist - the challenges you have to master - regardless of whether you use a 3rd party engine or a specially programmed one. and I find that exciting.

check out the new upcoming render engine from blender - eevee
or marmoset Toolbag - I know - i won´t start with this topic again - I'll stifle that at this point. but these are all just fucking tools.

nobody who hangs around on pouet - maybe with a few exceptions I wouldn't call anybody here a “lamer".

it is all technically interested people who are far above the average of those who call themselves "out there" computer experts.

so - i'm not afraid that any gamer lamers will show up here in the future and their video captures at demoparties.

and if that were the case - then this trend would end as quickly as it began with yelling “boring". where are my friends from metalvotze ?

;)

BUT - let´s be also respectful here: maybe ? just maybe - these gamer guys have skills - maybe in tactical thinking
or whatever - and this knowledge could be used for something better (in combination with a 3d guy or a coder or a sound guy) than shooting at enemies on the screen -
just a thought. combinating things is the key imo.

i am currently in the process of learning unity - with a little scripting, taking some baby steps here and there. combining it with my 3d stuff - and maybe i will do a demo with it in the future. and yes, i´ll write that i did it with that engine - and i´ll do it proudly.
on the other hand i have a lot of fun with my buddy pro from nuance on his own engine doing demos.

all said - my 2 cents

love, Ollie / cosmic of nuance

cosmicollie: sorry man, I think you misunderstood me here. It was just a reply to gloom after he was saying those effects (DOF, PBR, SS) should not be considered 'effects' in 2018. My point was to give him an example of offline render that does much better approximation of those effects (using unbiased monte-carlo raytracing) and convince him that real-time renderers are still no where near photorealism.
In fact in modern engines (UE/Unity) and in any demoscene tools, all those lighting effects are barely hacks that are approximating still only relatively small consequences of those complex lighting phenomena - usually the ones that are most visible to viewer, but still. The situation maybe will improve with real-time raytracing, but there are still many challenges there (even Smash actually had a talk about it) to make it fast/practical/noise-free etc...
I hope this clarifies my point.

I also really didn't mean no respect to Gloom, since I actually consider him to be really good at what he is doing, but maybe he should calm down too and stop undermining creators of 4k/64k intros and challenges in improving real-time rendering techniques.

everything's cool. dont worry.

i think the basic problem is simply a communication problem - between artists and programmers. and here there are many misunderstandings that are not even badly meant - everyone wants praise or respect. and i also said this to my friend pro a long time ago: "i can't really estimate what you are really doing and what's going on in your head and how much work it is e.g. to program a glow effect or a nice bokeh."

for me it is "actually a mouse click in after effects or in any tool".
shame on me. i know that now.

now that i´m sitting in front of unity - knowing nothing about
code - now - 5 years later i tell him: "dude, the more i get into this the more
i respect and understand what you are actually doing" - "sorry for being such a dick"

on the other hand there are so many opportunities that programmers might miss because they don't deal with 3dsmax or something similar. let's take a look at it this way:

good demos in the past - just eg. sanity / anarchy etc. etc. i think the coders
doing these works were also good gfx artists.

what smash at a revision party once told me was: he had to learn gfx. and how to do good design. i could imagine this was hard.

left brain hemisphere vs right brain hemisphere - logical thinking vs intuitive / creative thinking - these guys don´t like each other very much :)
so kudos for that.

when someone programs his own routines, experiments with his own engine - in his own code - then explores other tools like ue/unity/afterfx/blender/notch/blender/max whatever - and then comes back with a greater knowledge - and then combines this with his own skills or builds it into his own engine - then the works stand out and something new starts.

as said before: the treasure lies in connecting things

going to bed now - see you

cosmicollie: yup, I totally get you, so the communication between artists and programmers doesn't have to be that bad after-all :) g'night

It's ok, Tomas. Don't mind me and sorry for interrupting your thoughts here. Please continue.

Well that's not a too good example, really. The cracking/warez scene always had, and still has extremely strict rules about their releases.

Because this is the demoscene. I would think that was self-evident, but apparently it needs repeating.

I struggle with recognizing what argument you are trying to makehere: because real-time can never be photorealistic, we might as well just give it up? Because offline rendering yields technically better results, there's no point in doing these approximations in real-time demos? Did I at any point say that the primary goal of demos is to make things photorealistic? Of course I didn't, so you can put down that strawman right now - take a deep breath and read what I'm actually saying.

Regarding those techniques you listed, I was saying that on their own they don't constitute effects in the demsocene sense. More is required to make them interesting - you can't just slap DOF on a pretty object and say "There - that's my effect, LOOK AT IT! MARVEL AT MY ARTISTIC SKILLS!". Of course, you can do that, but it won't make for an interesting demo.

Jesus spoonfeeding Christ.

gloom: so we obviously have communication issues here. It was said before,but it just depends on perspective - for some slapping well-made DOF to a simple scene is already an 'effect', but of course I get your point that you may need a bit more, especially if you judge artistic side. But look, sometimes FLT demos have such simple 'effect' scenes e.g. aformentioned smoke scene in Wander - tell me here, smoke is kind of well-known effect, right? But still making it right and with enough detail/dynamic is still kind of cool, isn't it?

ton: I may get what's your angle, but your statement is oversimplifying the 'state of the art'. Yes, those effects are well understood from physical/optical perspective, and yes, some decent real-time approximations are well-known. However, it's still relatively easy to find a specific cases when those methods fail, e.g. people experiment with real-time raytracing for GI/PBR/etc... and you probably know that even for good anisotropic surface roughness model you need quite a few rays per fragment.

DOF is a great example actually.

10-15 years ago DOF was hard to do at all. putting DOF in your demo was a "wow" effect.
over time it became more and more well understood and hardware got better, and it got easier. most (3d-oriented) demos started to have DOF. it's now available as standard in many 3d engines (be they commercial, opensource, demoscene, hobbyist) - it's on the checklist of "stuff you implement when writing a 3d engine". the quality you can achieve in real-time is nowadays pretty good in the right circumstances.

there are a bunch of great quality docs and presentations about how to do a really nice DOF implementation on the internet. if you google "realtime depth of field" and spend half an hour reading, you're going to arrive at a decent "how to" or even some sourcecode.

DOF remains a useful design tool. but as a "stand out effect" it's passed into history. like how showing a spinning box used to be a demo effect too.. :)

Nice points being made. The realtime workflow is indeed pure gold. I couldn't have created all that recent spacepigs stuff without the interactivity provided by our tool. That includes both coding and design.

Another argument against demotools!

smash: I'd otherwise agree with that, but the implication is that something can be an effect only depending on the difficulty of implementation a particular platform. That's one perspective, but more so it comes down to presentation. I think it's good that it's become ubiquitous enough—just another camera parameter—that it's often used in a very transparent manner. But it still can be used as an effect, like you'd use it in actual cinematography (I'm kinda surprised nobody has really yet experimented with "creative" "bokeh shapes"...), like is the intention here. Similarly doing a swirly Helios-like bokeh, albeit technically optically bad bokeh is definitely something that could be used as an effect. I think it's good that most of the time DOF is not presented as an effect anymore, but just because it's become relatively pervasive shouldn't preclude that.

smash: we can of course nitpick forever,but... just curious.. are those popular DOF techniques finally handling transparent objects properly (in a coherent way, not as some hack on top)?

Gargaj, the tool in this case is mostly a live coding environment.

DOF is not so much an effect, I consider it a tool for guiding viewer attention. Same goes for FOV and all the basic camera stuff.

tomkh: no they aren't. and that's where it actually becomes an effect again when you take on those limitations. it's also something you can't do in any off the shelf engine, so it's a chance for custom implementations to shine.
the problem is of course that only a tiny % of the audience will have the technical knowledge to recognise your achievement. :)

jco: Most of the time, like it should be, but imagine a demo that showed everything through a super narrow depth of field, blending the image into an abstract mush. I.e. if you'd make a demo that looked something like this or this.

The term "effect" is pretty much all encompassing in demoscene context. So, rendering a rubber ducky with vertex phong shading once was an effect called "glenz vectors". Pretty wtf if you think about it.

noby: true that. I'd rather call that artistic expression using particular tools though.

To elaborate a bit further, I think to call something an "effect" is more applicable if you have to apply ingenious technical fakery to create certain visuals. Like those rubber cubes on a C64. Easy to do if you'd just render a mesh, and in that case not an "effect". But pulling this off with the by now well established tricks, I think effect is the correct term to use.

Here's a helpful graph as to how to distinguish between what's an effect and what's not:

lol. larusse fire clouds!

There, improved it.

STILL COUNTS.