Firstly, seasons greetings to everyone! Now, I could use a little advice.... I'm doing a commercial project. Basically, it's a boardgame, for which I've been commisioned to do the artwork using Poser for the characters and Bryce for the board itself. I've just completed the board, which consists of a 20' x 20' arial shot of a small town at night. It's a typical small town, containg park, river, bank, town hall, school etc etc. Everything is lit, lights in buildings shining through windows, streetlights etc, casting lovely beams of light and shadows everywhere. There's lots of trees, cars, and small details etc. Quite a complicated scene, and it's taken a month or so to put everything together. I have to now render this at print-quality. I've just spent Xmas with my in-laws, and thought that here was a great opportunity to do what I knew would be a fairly long render. Accordingly, I set the scene to render at 300dpi, producing a 144MB tiff file at the end, and then went off to enjoy my Xmas. I've just returned after 2 days, and found the render is only 1% complete!!!! I'm stunned. I knew the scene was complicated, and I've never done anything of this size before, but still... at this rate, I'm looking at a 200 day render time, and I just can't wait THAT long! Besides, I don't think my client would appreciate an invoice for 200 days of work LOL. In any case, the artwork has to go to the printers in early January! Does anyone have any advice for what I can do? Has anyone done anything similar? As long as I get a print quality render of the whole scene out of it, and within about a week, I don't care what I have to do. Perhaps I can import the Bryce file into another application for a faster render? Any and all suggestions gratefully received... If it's of any help, I'm using Bryce 6, on a 733 MHz G4 PowerMac, with 1.25GB RAM, and an ATI Radeon 9800 Pro 256MB graphics card. Many thanks, Rob

This is an interesting problem. I just spent Christmas Eve rendering a scene that crashed Bryce 6 several times and gave me "out of memory" errors during saving.

Anyway, I am curious. How about, if possible, segment the scene into various parts and render those. Then postwork them back together.

Although my issue isn't the same, it does share some similarities: problem getting it done. I wish you luck, and if you do find a solution, please pass it along!

I'd 1st of all use ranged fall-off on lights. After a certain distance lights from small houses don't matter for the rest of the scene (especially if theres no indirect light, like any bryce scene that doesnt use TA). It'll possibly save u lotsa time. Also volumetric light effects would be more efficiently done in post.
I'd also turn of TIR, prolly no use for it in this scene, and maybe check out if u can limit reflective surfaces. Raydepth can be kept low if theres no prominent reflective or refractive elements in the scene also.
And if ur rendering for 300dpi, theres no use for insane AA settings either, so dont let urself get swept away with those either. Normal AA is prolly gunna do just fine.

Out of curiosity, how big's your Bryce scene file on hdd? And do you hear your HDD grinding away constantly while its rendering?

I once had to render something that was 36"x20" @ 300dpi and I ended up rendering it in quarters and then putting them together in photoshop - it didn't make it go any faster, just broke it up into bite size chunks.

Rayraz has made a pretty decent list, Also, I'd look at eliminating any soft shadow settings - they can eat render time up as well (basicaly simplify your light setup if you can).

200 days sound crazy, even for a highly complex scene.
Soft shadows(as Flak pointed out), volumetric lights, DOF, highres image based textures & transparency masks, glass and reflection all eat render time, so try applying what you can with postwork. It can save you weeks (or in this case, months) of rendering.

Perhaps you should also consider upgrading your machine if you´re going for heavy scenery. 3D require a lot of power from the hardware, and unfortunatly it doesn´t sound like your CPU and the amount of RAM will be enough if you need to speed things up.

One other thing i would consider is the resolution. 300dpi for a normal sized boardgame sounds like overkill to me. For most printing, 72 dpi and render resolutions around 5-6000 pixels is usually more than enough.
Unless it´s a very large boardgame. :)

Good luck.

Hi, thanks for your responses so far... to answer a few points raised: 200 days is a guess based solely on the fact that it's taken 2 days to render 1%, nothing more. But if rendering continues at the exactly the same rate, then it would take that long. Whether it really will take that long, I've no way of telling at this point. I've used range fall-off on all the lights; the scene is intended to look as natural as possible. No volumetric lights, or reflective surfaces. Although there's a small lake and a river in the scene, the light going across the water surface is at a very low angle, and there's little (if any) reflection. There's glass, of course, in all the windows in the buildings and cars, but most of these are so small, and viewed from directly above, that they shouldn't really affect things too much. I suppose I COULD go back into my file, ungroup everything, tab thru all the hundreds and hundreds of elements, deleting all the little panes of glass.... but that would probably take as long as the rendering. I'm using normal settings, not superfine rendering. The Bryce file is 293.2MB. Doesn't seem to significantly slow down my hard drive (I can post here fine, and surf OK, and right at this moment I also have Photoshop open too). Rendering in segments is an interesting idea, I hadn't thought of that. Will give this some serious consideration. It absolutely has to be 300dpi. 72dpi works only for screen renders or low-quality small images. It won't work for professional print jobs. 300dpi is the minimum resolution I can use. And finally, yes, I'd love to upgrade my machine; one of those latest shiny new G5 dual core processors would do for a start if anyone feels like buying me a slightly late Christmas present:-) But I was actually hoping to use the fee from this job to pay for one, can't afford it right now! Thanks for your ideas. Please keep 'em coming! Rob

Some advice i can give you with bryce renders that i have picked up from this forum and trying things out myself.

Poser figures:: if you can't see things like teeth, tongue, legs, feet, shoes or any parts you can't see select them and delete them in bryce.

Lights:: you will find the more lights you use the longer your render will take.

If you have a lot of trees in the scene try replacing them with 2d ones, trees cause a render to take longer.

Render settings:: There is no need to use anything Super with an RPP(rays per pixel) setting of 4-16, if you need to use the Premium setting you don't need RPP above 16 and remember every one of those Premium Effects you tick adds to render times.

While having a render going you would be best not using other apps or have anything running in the background like Anti virus, firewall or anyother app that will use ram/cpu.

Quote - It absolutely has to be 300dpi. 72dpi works only for screen renders or low-quality small images. It won't work for professional print jobs.

 
Unfortunatly i´ll need to disagree on this. If the client demands 300 dpi, there´s no question. 
But i actually work in a printshop, printing out high quality art posters among other things, and 72 dpi works perfectly fine in most cases IF you render out large. 
You simply wont see the difference in the end result.

Unfortunatly 300dpi have become something of a mantra in the business and everyone wants it, but there are actually very few cases you really need it for printing purposes.

Nuff about that...

Are you using a lot of highres textures? If you do, try changing some of them and see how much difference these make. Your windows shouldn´t make that much difference unless they have a very high reflection value.

If you have any visible lights(render options in the Light Lab editor) with soft shadows, that will kill a render time, even if its one light...

It seems to me that it wasn't the client that demanded 300dpi, but a target you set yourself - correct me if I'm wrong - If it was indeed the client, take him behind the bike-sheds and 'educate' him......................

Hi again, Some progress: today it's reached 2%! At least that's speeded up a bit:-) 2D trees: I see exactly what you're saying, and would normally have done this, but in this instance it's a view from above, so the trees have to look OK from above and also cast a normal shadow on the ground. That's hard to do with 2D trees. No visible lights in the scene: there should be, if I was going for absolute realism, but I already cheated on that one. Rochr, I am VERY interested in your response: and a little voice at the back of my brain is whispering that this may indeed be the solution to my problem. Yes, 300dpi has always been used as the minimum, certainly since I've been in the business, working first in trad media, now digital, over 20 years. Digital printing methods have changed that. It was the client who asked for a 300dpi render, but this isn't set in stone: what he wants at the end is a high quality 20 inch square gameboard, and I guess he doesn't care (or needs to know) what hoops have to be jumped through in order to get it. I suspect, like many people, he uses the term "hi-res" to mean "high quality", when, as you point out, they need not be the same thing. So, may I ask you to expand further on this? If I need to print a 20 inch square good quality image, what size would be best to do the render at, at 72dpi? I eagerly await your reply! many thanks, Rob.

Well, since you´re in the business as well, you probably know this already. But anyway, for a standard 70x100 cm high quality poster, a 4200x6000 uncompressed TIF in 72 dpi is more than enough. Naturally, the end result also depend on the printer and material, but  you should be able to go for 5000x5000 without problems.
If you want to be on the safe side, up the resolution with 1000 pixels on each side. It will still save you a lot of time compared to a 300 dpi render.

Hi Rochr, Many thanks for that. When I say I'm "in the business", I meant producing art, as opposed to printing it. I've never in 20 years been asked for anything less than 300dpi for a print job. A screen job has always been at 72dpi. Maybe it's just the clients and printers I have worked with, and that's just one of those things... Anyway, what you say makes perfect sense; render it large, at 72dpi, then when it's reduced for printing, there is no apparent loss of quality. I'm definitely going to do this, and I thank you for sharing your wisdom! My best wishes to you and yours for 2007! Rob.

No problem, just glad i can help. Hope you can speed things up. 
Just remember the great tips you got from the guys here as well. Each and every one of them will save you countless hours of render time. :)

These forums are always my first port of call whenever I've had any kind of issue. It's such a fantastic resource, and I've never failed to find some kind of solution, even if it wasn't the one I was expecting. I salute each and every one of you. A happy, peaceful, and prosperous New Year to you all. Rob

Quote - I've used range fall-off on all the lights; the scene is intended to look as natural as possible.

I'd doublecheck if they're reduced to as little range as possible. (aka, if u only realistically get stuff in a radius of 50 bryce units lit theres no use using a limit of 200 units)

Quote - No volumetric lights, or reflective surfaces. Although there's a small lake and a river in the scene, the light going across the water surface is at a very low angle, and there's little (if any) reflection.

Sounds good. Also no volumetric world i assume?

Quote - There's glass, of course, in all the windows in the buildings and cars, but most of these are so small, and viewed from directly above, that they shouldn't really affect things too much.

It's a top view right? in that case those glass things should be barely visible. I'd either remove them or, if u really really have to, use something with plain transparency (u know the kind that doesnt refract). The surfaces might not have many visible pixels but if theres lights behind the glass windows the glass will still slow down the rendering of those lights.

Quote - I suppose I COULD go back into my file, ungroup everything, tab thru all the hundreds and hundreds of elements, deleting all the little panes of glass....

Check if ur grass plane texture can be made as low-res as possible. A plane that shows as a few pixels doesnt need a map of 1024x1024 for instance. This might save u some memory usage which will possibly speed up the scene

Quote - but that would probably take as long as the rendering.

Are u sure? If u think u can optimize things significantly, i'd say its worth a shot! And if u cant avoid a render-time debacle, make sure u tactically inform your client before its absolutely totally last-minute

Quote - I'm using normal settings, not superfine rendering. The Bryce file is 293.2MB. Doesn't seem to significantly slow down my hard drive (I can post here fine, and surf OK, and right at this moment I also have Photoshop open too). Rendering in segments is an interesting idea, I hadn't thought of that. Will give this some serious consideration.

If you're gunna render in segments anyways, try spliting your scene up to multiple scenefiles each containing only area's of your total scene that cover mainly the segment you're gunna render. If that's possible for your scene-setup. (even grand master rochr seems to use multiple seperate renders of different parts of scenes, it's really an efficient trick!)

Quote - It absolutely has to be 300dpi. 72dpi works only for screen renders or low-quality small images. It won't work for professional print jobs. 300dpi is the minimum resolution I can use.

For a full-page professional magazine print in magazines such as 3dworld 2500 to 3000 pixels wide or high generally suffice. thats for almost 30cm of image width or height. I wonder if that's a decent guideline?

Quote - And finally, yes, I'd love to upgrade my machine; one of those latest shiny new G5 dual core processors would do for a start if anyone feels like buying me a slightly late Christmas present:-) But I was actually hoping to use the fee from this job to pay for one, can't afford it right now! Thanks for your ideas. Please keep 'em coming! Rob

Dont worry, it's a know problem Takes lotsa saving before u can buy the fast system. But once u got it, it's so worth it! haha. Anyways... if ur lookin into a carreer of professional 3d graphics it might be a good idea to also slowly start looking into some other software that renders faster. Especially if you love to do highly complicated scenes. Faster renders are just that little bit less stressfull when workin for a client u know :biggrin:

Quote - ...it might be a good idea to also slowly start looking into some other software that renders faster. Especially if you love to do highly complicated scenes.

That´s actually a tough choice. What bryce lack in speed, it compensate with beeing able to handle extreme polycounts very well, which would be necessary for those complex scenes.

First thing I'm trying is the large render at 72dpi. If that proves to be still too slow (and I'll know by tomorrow morning), then I'll try doing it in segments. And then I'll just keep adding each tip here until I get a workable solution. And RayRaz, I don't necessarily love doing complicated scenes; it's just sometimes I have to:-) Anyway, although I work in 3D, all of the work I produce is for eventual 2D output, so I am able to cheat an awful lot in post production. Rob.

Quote - > Quote - ...it might be a good idea to also slowly start looking into some other software that renders faster. Especially if you love to do highly complicated scenes.

That´s actually a tough choice. What bryce lack in speed, it compensate with beeing able to handle extreme polycounts very well, which would be necessary for those complex scenes.

oh yea ur right about that one :-s damn.. i almost forgot about that haha. ur absolutely right! though there's a few memory optimizing functions in mental ray that i know of (u do need to set falloff limits on ur FG to make it actually work in indirect light calculations) but u do tend to have probs with polycount much sooner then with bryce...

OK, Just thought I'd update you on my progress. I've implemented all the changes that have been suggested so far. I've made some significant improvements, but not significant enough. What I've been doing is implementing a change, rendering 1%, seeing how long that 1% took to render, and then simply multiplying that by 100 to gauge a very rough estimate of how long the whole thing might take. Now, I know that's probably not the most accurate way of doing it, but I don't know any other way. Currently, it's taking about 12 hours to render 1%. This means it will probably take approx 1200 hours, or 50 days, to render 100%. While that is a HUGE improvement on 200 days, it's still not fast enough. I've got about 21 days before this has to go to print, and I need to really slash this render time further. Does anyone know of any further thing I can try? I have heard of a little device called a RayBox (I think), which is sort of like an external hard drive, only it's solely devoted to 3D rendering: it simply plugs into your machine, and off you go. It doesn't feed off your machine, so you're free to get on with anything else, while this takes care of the rendering. Trouble is, I can't find out anything about it, no pricing or availability. Anyone heard of this? I've thought about the possibility of rendering in stages: i.e. remove all trees for example, then render. Then render a version with just the trees, and then one with just the buildings, and so on, putting it all together in Photoshop. How does this sound? It's a lot of postwork, but if it can save me on render time, it might be worth it.... Once again, all and any suggestions are welcome. Many thanks, Rob.

One other thing: from digging through the forums, several people seem to be saying that Bryce 6 is slower in render time than previous versions of Bryce. Is this true? Could I open this file in Bryce 5.5 (or even 4) and render faster? What do people think?

Rochr, your example of a 4200x6000 image being used for a 70cm by 100cm poster works out at approx 150 dpi, not 72dpi, which is I believe, fairly usual for poster size, and so should be acceptable to look at. If you printed that image at 72dpi, it would end up around 211cm high. 

At the end of the day, (for example) a 4000x3000 at 72dpi image has the same number of pixels as a 4000x3000 image at 300dpi, the only difference is the output size of the print, as the dpi (which really should be ppi) is telling the printer to fit so many pixels into a certain area.

There is another option, and that is plugins (ie Genuine Fractals, or SizeFixer) for Photoshop that can enlarge pictures with minimal quality loss, so doing a smaller render size and upsizing could be quicker too. Not cheap, but affordable, especially if you need the time.

http://www.fixerlabs.com/New_Website/pages/sizefixer.htm

http://www.ononesoftware.com/detail.php?prodLine_id=2

Another option I suppose is do you know anyone who has a few computers you could use for a Bryce Lightning network render?

Yeah, agreed, even a 4200x6000 can be overkill, but normally i rather render stuff out larger than necessary. That way i can up the size of the print if needed, without having to re-render the stuff. 
Safety precaution. :).

I've seen your pictures, you need them that big just to fit the details in ! :)

Very interesting. So, are you saying that I could go buy SizeFixer, render out at a smaller size (for example 2000 x 2000) and then boost it up to the required size with no loss of quality? This could be worth doing if the end result still looks good.

That is what the makers say, though not having used them myself I can't vouch for them. 

Fixerlabs has a service that's reasonably priced where they will upsample an image for you, and give a coupon so you can get the cost back if you purchase SizeFixer. That might be worth trying out.

The Genuine Fractals site has a fully functioning demo download that apparently lets you do 20 images before ceasing to work so that you can evaluate it, here:  http://www.ononesoftware.com/download.php

well, risizing apps are never gunna perform miracles, but they're better then nothing.

Well. from the demos the results look pretty amazing. And although I'm not dumb enough to expect the results to be BETTER , if they are at least AS GOOD AS, then it might be the only way forward I have. My deadline is so my client can make a prototype to take to a European toyfair, to try and get manufacturers interested. So it doesn't have to be absolutely perfect for the prototype. We'd have more time after the toyfair for tweaks and redos.

Just be careful when using resizing tools. You may end up with an end result looking like something shot with a cheap digital camera. These softwares have a tendency to take away the sharpness in an image.

I'm hoping to still render out larger than the actual printed size (20" x 20"). If I render out at about 30" x 30", then boost the size just a bit more with the resizing plugin, then when the whole thing is reduced back down to 20" x 20" for printing, it should pass muster. At least, that's the plan:-) Anyway, for the toyfair mockup, which is my priority right now, it should be good enough. Rob

yea sounds like decent trouble-shootin there

I get lots of images printed for exhibition stands, sometimes as large as 3m x 4m in size. My standard procedure is to supply the printer with artwork 25% of the finished size at 300dpi, the printers do whatever it is they do? to make crystal clear prints from the file. It might be worth while running a test render to see how much this reduces the render time. If it works in your favor act all innocent with the client and explain to them the whys and wherefores of the situation and that you just assumed it is what they meant!

You might just check with the printers first though.

Good luck!

I'd say be carefull what u tell ur clients... It's often best not to twist reality too much.

Where is reality being twisted?

The part where u render at 25% and let the printer scaled up while u know ur client requested X" wide and Y" high at 300dpi?