Eric

"unprintable" Define unprintable. You have to determine your final presentation method, viewing distance and acceptable print quality. Just because the print company tells you they need their images submitted at 300dpi, that is just a standard number they throw out there. On that note, "DPI" being used as an image size is laughable because DPI refers to Dots of ink Per Inch, which is extremely low. We use "PPI" (Pixels Per Inch). Anyway, that's off topic. I read a book years ago that showed the difference between two identical images printed on the same page. One of the images was 300ppi, the other was 200ppi. You could not tell a difference, at all. Now, if the page was printed on a museum quality printer, you could probably tell the difference. But only if you were less than a foot away. After I read that book, I started rendering all of my images out at 200ppi. Once I started rendering images out for 24x36" prints, I realized my computer wasn't fast enough to render out the quality of images I was after, so I stepped down to 150ppi. Guess what - NO ONE NOTICED the reduction in print quality. I rendered an image for a billboard once. The final image I submitted to the print company was something like 50-75ppi. It looked fantastic from my car zipping down the highway. With so many presentations being made digitally nowadays, I've been rendering out images at 1920x1080. As 4k screens become more common, I'll probably have to increase my output size. If I know up-front I'll be printing the images, I'll adjust the output accordingly, but 300ppi is never my goal. I'm quite happy with 200ppi. Especially since most of my prints are on an 11x17 laser, which can't compete with the DPI output of a good inkjet printer. My .02 cents. Edited to add - the 8000x6000 image size requested in the original post may be dictated by large format prints they display on their construction sites. If they have a print company that they regularly send their work to, the print company may have determined that resolution works best for their particular final print sizes and gave them that size to use as a standard when hiring their CG resource(s). I wouldn't necessarily take their request to mean it's the new industry standard.

I'm running the subscription version of 3ds Max 2016 (just upgraded from 2015 a couple weeks ago). I frequently use Batch Render, and when I went to create a new Scene State with Max 2016 with a few layers selectively hidden for particular views, the Scene State menu item was missing from the right-click quad menu. Clicking on Tools>Manage Scene States does absolutely nothing (no dialog box pops up). Any suggestions? I hope I don't have to repair/reinstall since it requires I get my IT involved. Hoping it's just something I'm overlooking.

Sounds like a fun project! Here's an article posted here a few years ago. http://www.cgarchitect.com/2012/06/building-an-accurate-3d-model-of-london

I don't know what I'm talking about. That being said, what if you tried locating your working projects to a sync'd location, like a Google Drive folder on your desktop so all the other computers are sync'ing that folder locally. Then, tell Vray to NOT "Transfer missing assets" in the Distributed Rendering Settings dialog box.

Hopefully this doesn't reply twice, but my previous post wasn't showing up. Under your "Indirect illumination (GI)" tab in the render settings, reduce the "Saturation" value from 1.0 to something like .25, or even 0.

My IT is concerned about buying a graphics card that is not on the Autodesk "Certified Hardware" page. I'm leaning towards a pair of GeForce GTX 980 cards for GPU rendering. If it works out (GPU), I plan to add matching cards to my old computer (the one I'm currently using) and my render slave. I run AutoCad, Revit, 3ds Max and Photoshop. Is there a problem with going with a card that's not on the Certified Cards list? Can anyone tell me why, or why not? If the card is not on that list, will it show up in the GPU card list when I am in the MR iray activeshade dialog box? Or do I have to somehow force Max to recognize the card...?

My IT guy is about to order a new computer for me, but he wants to go with an AMD graphics card. I'm pushing for the nVidia since it's required for iray. Can anyone tell me why I should go with one over the other, besides iray? I'm really wanting to try GPU rendering, or at least try to start using Active Shade for testing. I used V-Ray RT when it first came out, but didn't really have the hardware to support it. I don't want to make that mistake again.

Nope - never got any additional information. It seems like I remember seeing memory usage numbers displayed in the log window for the Production renderer, but I'm not sure how accurate, and at what stage that number is pulled. Plus, it would include geometry that is calculated on-the-fly, like V-Ray Planes, Fur poly's, etc., which GPU ignores anyway.

This is something I recently asked on the Chaosgroup forum. http://www.chaosgroup.com/forums/vbulletin/showthread.php?68827-How-does-the-GPU-memory-and-scene-size-relate

Maybe I'm missing something, but it seems the easiest thing would simply be creative use of the opacity map and blended/composite materials? Here's a quick test using nothing more than a metallic material with a checker map applied to the opacity channel with a fresnel falloff.

I'll give you a very recent example... I have a restaurant model that was taking my 3 computers about 24 hours to render a single view. I have 12 views per model, and 4 models, so 48 images total that I'm generating over the course of two months. My computers are now 2-3 years old Intel Xeon quad-core single processor Dell Precision T3500 workstations. I've been using RenderCore out of Los Angeles, CA and have been extremely happy with them. For the same images, their computers can render the images in 4-16 hours (depending on the view). Their computer specs are listed on their website if you're curious. They put one computer on each image. They can also do split-frame, but you need to pre-calculate the lighting or you'll get noticeable banding. Last night I submitted a batch of 12 views (one model), and the last image just finished a short while ago. My total bill to render 12 images was $700. To break that down, it took a total of 125 hours, and they bill each hour of processing time at $5.60 per hour. So, on average, it cost less than $60 to render a single view that took their computer approximately 10.4 hours to render. Had I rendered those images internally, I'd have needed over 5 days straight of render time to meet the deadline. And that was for 1 of 4 similar models. Oh yeah - if you have an idea how long the images will take to render, you can pre-pay and get nice discounts on the per-hour price.

I'm sure you're seen the V-Ray help file, but here is a link to the V-Ray Sun page. http://www.spot3d.com/vray/help/200R1/examples_vraysun_sky.htm The image you attached has a lot of compression artifacts. It's hard to see what you're trying to point out. The shadows look like they are crisp in my opinion, but the jpeg compression is blurring everything in general. If you're after soft shadows, you need to use the V-Ray Sun and increase the Size Multiplier till you're happy with it.

When we first setup our farm, we had similar issues. Our first solution was to place all the maps on our server (not just a render server, but our actual file server for the whole office). We then installed "Allway Sync" on each of our render nodes and told them to synchronize the maps folder on our server with each of the node's "C:\Maps" folder. Moments after anything was added to the server's Maps folder, it was automatically copied to each of the render nodes. That worked okay, but it was silly to have duplicate copies on all the nodes. We then switched to saving all the maps to one computer and using proper UNC paths. The benefit here is changes to materials happen instantly - you don't have to wait for a synchronization program to catch up and distribute the files across the nodes before you can click 'render'. I do most of my modeling in Sketchup, then import the file into Max. Any time I make a change to the model and have to re-import an element, I have to open the Asset Manager and fix the path to the material so it uses UNC paths. For what it's worth, I've never been able to simply tell the Asset Manager to resolve the paths to UNC - it doesn't actually do anything for me. I have to manually select a group of maps that are calling the same folder, then fix the path manually.

Michael - when I initially replied to you, I hadn't noticed you referred to one of my old portfolio renderings. Sorry about that. To give you some info about that particular scene... All of the landscaping is 3D models from one of the Evermotion libraries, aside from the two foreground trees which are photographs inserted via Photoshop. The buildings on the left and right - those are from another library I bought a LONG time ago - I can't even remember the name of the library, but it's filled with stuff like (obviously) buildings, street lights, benches, etc. - street clutter stuff. It's all in 3D. Always populate the horizon line with something - even if it's a forest... The cars driving down there street - I enabled motion blur and animated the cars driving down the street and positioned them where I thought they looked good, then rendered that one frame for my still. I probably could have added several more animated vehicles. The shadow on the concrete in the foreground - that was a duplicate of one of those trees which I filled with black, blurred, then distorted to make it look 'right' with the scene. The dark streaks on the street under the cars - those were two horizontal lines I painted with a soft brush in Photoshop on their own layer, then I distorted it to give it the correct perspective to line up with the street - much easier to do this method than to try to paint them straight onto the street and get the vanishing lines to match. Although, as I look at it now, I see I could have tried a little harder... The Audi on the right is a photograph I found on Google, then masked out and saved to my library - that was from before I had a Dosch library of 3D vehicles. I added a little white glow to the headlights. The sky was a Photoshop gradient, using the alpha channel of the .tiff or .tga output from Max. The corners have a slight vignette applied for a 'photographic' feel. The concrete has a black gradient that comes up from the bottom of the image to darken the foreground - I reduced the opacity and played with the layer's blend mode to get an effect I was happy with. There is a slight noise added to the image for a film grain feel. The windows were rendered out of Max in a way that allowed me to select them as a mask (applied a self-illumination with no other lights or GI turned on so they rendered as a solid color I could select in Photoshop - much easier to do via V-Ray, but I've never actually played with that feature yet). I then applied a slight green tint to them per the client's request. Anyway, point being - you can get the model to render that way straight out of Max if you really wanted it to. But, it's much easier (and usually more cost effective) to get it 95% there, then finish it up in Photoshop. This guy taught me that... http://www.cgarchitect.com/2001/12/interview-with-ernest-burden-iii-of-oreally-inc http://www.acmedigital.com/tutorial/tutorialLS-WC.html This guy has some nice tutorials too... http://alexhogrefe.squarespace.com/tutorials/

Are you allowed to post your image so we can see it? How are you lighting your scene? Are you using the V-Ray Sun and its environment/sky light? Are you keeping the sun relatively low in the sky, or high and bright? Low is better. Is your sun favoring one side of the building more than the other? Are you using the V-Ray physical camera? Are the renderings from the other 3D computer something they sent for marketing use? If so, I don't see any reason you can't at least link to their website so we can see their work for comparison.