I was delighted by it. Finally it worked the way it should: be ware though: have the light normals pointing the right way … also I searched a LOT for the show normals/flip normals that used to be on the f9 screen … no it’s there only if you’re in edit mode smart.
THE GOOD & EXCITING STUFF
Anyway … my happy experiment was with 3 planes aranged as if they were the floor left wall & ceiling of a room, and another small plane as a light source pointing maily towards the floor. I was delighted to see light go to the ceiling. I thought to myself that Blender had finally got rendering right … and maybe it would be good enough for me not to install anything else but Blender (I’d love it if yafray or whatever came in the package … as rather than risk going into all the trouble of having to configure scripts, python paths & packages … I’d rather use a less features thing … and this new radiosity might just be it … however …)
The bad stuff
I tried to make a sky dome & have that iluminate some bananas I made a while ago … they came out like a skin disease. Thing is that it seems the new sistem still uses vertex colours … and thus it wiped my old painting of the bananas which made them yellow & black pleasantly colourfull.
Another sad experience was in an attempt to get soft shadows: placed a cube and a small light emmiting plane above it and a floor … it didn’t cast any shadow!!! Just that the floor wasn’t white, but grey-ish. I guess it must be working per polygon … so I subdivided it … and yes … the shadow started formign … however there were 2 flaws: being dependent on the subdivision and the subdivision not being exactly suited for the shadow … it was a non-acurate shadow … and even more noticeably … a chessboard type pattern stared forming as it seems along the lines of the subdivision the surface was darker than the rest.
Yes, subdivision is required if you want more detail in the radiosity solution (unlike the other method, it isn’t subdivided when rendered). As for your other problems, try to avoid faces that are too small, and it isn’t as much of an issue if you hemirez (in the radiosity buttons, similar location to world and texture buttons) as high at will go (which will make things slower, but…).
yeah, its pretty fun, but, you must take some care to make it work.
I found due to the various vagaries of the system, I had to do closeups in totally different setups from my long shots.
The good? It looks nice, you can setup very quickly if you’re lucky, and you can get exptremely fast renders (again, if your scene works for it)
I used it in PDS https://blenderartists.org/forum/viewtopic.php?t=16810&highlight=
alongside regular lamps.
rendertimes for a lot of shots were under a minute per frame at 800 by 450
yep ;-) I'm still in luv with blender ... it's just that I always hit my enthusiasm against this guy I might start a 3d company with who points out to me the flaws of stuff done with blender ... and it really pains me to admit he's often right ... like for example .. maybe it's the people standars and time .. or maybe blender doesn' scale very well extremely complex geometry ... but I see stuff done in blender ... like say Andy's mindfields movie at www.blender3d.org ... which artistically I think it's awesome ... well thought out, cool mosic, nice plot ... but in details like say how those pillars look .. it lacks certain standards expected of today's 3d stuff .. like extreme details like cracks & stuff, tiny bumps that really are bumps .. or something .. I have yet to figure out what it is that's not good enough exactly (another example that comes to mind is the house in the 2003 conference intro movie) ... looks very low poly. I'm wondering if it's blender's fault or if we can do more. Anyway, I'd like Blender to get better (like say joining holes in geometry, face & line modelling modes ...) ... hopefully even to the point where I'll be able to use it on a comercial scale. I'd love to work in blender comercially ;-) ... but until I figure out how ... I should probably plan for others and watch with interest Blender's progresses (eg. recently I was really impressed by a finished project when for the first time I saw in Blender a plant scene ... and it was cool ... yes, it was quite tiny ... but still ...)
…or just use yafray. Rendering limitations need not be an issue if you have external renderer support ( MAX is a good example of this). We’ve got Yafray and RIB via scripts with I believe ‘integrated’ export coming soon.
I don’t remember which site site it was I found on (maybe 3dcafe?) … but there was something like “top 10 good things to do in a demo reel” (the bad ones were funny imo) … and unfortunatelly a clear idea I got from there is a couple of things that I don’t see as very obvious in blender … or rather I see as quite hard to get right … maybe it’d be interesting if the people who knew how to get them right in Blender would write a compilation about the stuff. Anyway … my impression of this stuff was mainly about natural phenomenon:
a) water
b) fire
c) wind
d) cloth
e) colisions/bouncing/physics
f) vegetation painting/painting of pseudo random stuff like say grass, cracks on an object …
it's quite probable that my knowledge is limited ... but my thoughts on these subjects and how/if they can be done with blender are:
a) I have no idea how it could be done in Blender (I higly doublt it would look right with particles … and it would be a pain to mesh some water flowing out of a bottle)
b) I’ve tried it … didn’t come out very well … a friend tried it … looked good in stills … not very nice in animation … certanly not up close … generally speaking I don’t know how you’d make it looking realistically
c) I’m not talking aobut just simulating it … but more like putting in “wind” in your scene and having leaves falling nicelly & stuff
d) I’ve got no idea how it could be done in Blender. Afaik it can’t.
e) Two things would be handy here: colision when modelling … so taht you can then make a scene and just scatter some stuff around and it would stop falling or flying or whatever when it hits geometry (nice way of filling a treasure chest without worying about hte gems intersecting)
f) Like I said … the problem I can’t solve in Blender is achieving extreme and pleasant detail. Eg. … say I have a surface … I might want to paint cracks on it … or maybe grass so it patches up nicelly and it’s not in one big thing
well ... this is my wishlist ... I assume everybdoy has their owns :)
PS: not to be forgotten: would love integrated yafray or renderman caps … possibly also with a ‘quick render’ region a la XSI
I’d have to disagree, at least with your d) e) and f) anyway. A demo reel is about showing off your skill as an artist. If you’re aiming for modelling, then having a good sense of form, propotions, detail, technique, or for animation, a good sense of timing, weighting, storytelling, or for lighting/TD, colour theory, composition, textures, etc. Things like physics and collisions don’t show off your skill - especially when most software these days has it all built in at the touch of a button, or with cookie-cutter plugins. Anyone can make a nice garden scene with maya paint effects, and anyone can put a bunch of objects in mid air and click on ‘rigid body dynamics’ etc.
Sure, those things are useful for production, but a demo reel is about the fundamentals - showing off your own talents, not your software.
i wish to use blender commercially (architectural visualization), but it’s hard to use it professionaly because of at “commercial” you haven’t enough time to make things looking right in blender. An artistic aproach can always be done with blender (as my old escalera image that simulates true radiosity with only lamps) but in real world you have no time to spend on putting hundreds of lights all over a familiar house to get some true radiosity feeling…
about blender radiosity… funny to experiment and make artistic approachs to reality… but you can’t belive it will cast a real looking ilumination using the real light scheme… have you worked with max integrated radiosity? first of all the scene MUST have real world units because of the real decay (i think it’s the word) of the true light, Real distances are a must to get good looking illumination in max with real world light schemes.
and… what happens with blender radiosity… distances and deacays aren’t so important? i don’t know, but max radiosity works and bleder one is a toy
your points are very good .. but they're a bit too idealistic for this world. I mean: yes, it would be nice if somebody for example hired me in Photoshop because seeing what kind of stuff I can do in Gimp they reason that a person who can do that stuff will learn fast and work good in Photoshop too ... but I don't think generally employers think like that .. but rather more on the lines: if this is how your demo looks .. this is the best you can do ... if your demo has this crappy water ... then I don't want you in my company because I just saw the demo of this other artist who had awesome water. My point being: I'd like to have you as a boss as far as this reasoning goes .. but in real-life I think fancy stuff *is* needed. Of course this is my opinion, that is yours .. and we don't want to go fighting over it :)
After all ... we both hale the allmighty god Blender :)
aqa: though an objective viewer might be able to criticize blender a lot compared to the best tools out there … we’re not supposed to be so objective we’re a bit more blender biased … so despite all it’s problems … let’s cross our fingers and hope it’ll be better in the near future
PS: if one was to render a scene in yafray … would the radiosity come out perfect? even without high detail mesh ?
This is all well and good, but you do have to make the consolation that Blender is an artist’s tool first. Accurately depicting the real world comes second to conveying the message. That’s why Blender doesn’t have a lot of the tools commonly found in many CAD programs. This isn’t a bad thing at all, and while the Blender base may be extended to include these tools to accomodate more users (also a good thing), the artistic heart is, IMO, core to giving this app the personality it has, and that should not be forgotten. Using Blender for highly accurate architectural visualization is like using a pallette knife to draw a blueprint. Sure, it can be done, but it’s not where the tool’s strengths are, so it’s going to be more difficult.
A toy? Hardly. A great tool that has room for extension? By all means.
Not fighting, but if your demo reel has water and fire and character animation and landscapes and… you get the point… then you probably need to redo your reel. Your reel should be targeted for what part of the process you’re interested in. If you’re main interest lies in character animation or modeling, then you might not even have to show fire and water. However, if your major thrust is in effects, then you better have the best fire, water, grass, and cloth on the planet, and modeling and animation would play smaller parts on your reel. Hell, if your interests are mainly centered around concept and pre-production, you’ll have mostly traditional sketches, sculptures, and storyboards with very little of either of the things above… and the stuff you have that is there might actually be created in 3D by someone else (be sure to make this known in the reel).
It’s all about what part of the field your mostly interested in, because, to be honest, unless you’re in a super small company, you’ll probably only focus on one or two things (for example, I know someone who spent 3 years in a large animation company animating fire, and nothing else). That’s the reality of things.
There are more realistic approaches to using Radiosity rendering via Blender. You can learn to use Yafray and PovRay Plugins which are both capable of allowing you to do the highest of professional 3d rendering. And I said learn because Blender is worth learning even if you own other 3d apps.
Even 3d Max users use other render engines like Vray and FinalRender. Yes 3d Max now has a great internal rendering system but that does not stop people from looking elsewhere.
The main problem with raytracing and production is speed. And professional 3d artist often toss aside the fancy bells and whistles of raytracing in real productions. Often effects like caustics and GI are faked.
And Blenders rendering system , if properly understood by the user, can deliver products that can match any other 3d application. So don’t be so quick to blame the 3d tool. 3d in general has a learning curve. Always be prepared to take out time to learn. Then later on when you want to do a professional project you won’t have many disappointments.
I would like mostly cloth and liquid. Fire can already be done pretty well with particles and liquid with metaballs, but not very efficiently. (Needs LOTSA processing power…)
smart.
imo) … and unfortunatelly a clear idea I got from there is a couple of things that I don’t see as very obvious in blender … or rather I see as quite hard to get right … maybe it’d be interesting if the people who knew how to get them right in Blender would write a compilation about the stuff. Anyway … my impression of this stuff was mainly about natural phenomenon: