(help) I need VERY high resolution and DPI for large high quality art prints
RexRed
Posts: 1,225
I have created a scene in Bryce3D and then I added a very large 2D picture object. Resolution dimensions 11454 x 8729.
I have 8GB of RAM. I used a program to allow Bryce to use more RAM.
I am using a 3/4 ratio and 600 DPI.
I used render to disk and set the DPI to 600 and the resolution to 41454 x 31592 Bryce said "not enough memory". I am wondering if the memory expansions app change is really working.
So I lowered the resolution and then part way through Bryce crashed and gave me an error message saying there were faults in some DLL's and a few other files.
So I lowered it even more to about 10000 x ... it seems to be rendering it but this is not what my printer person needs.
What is the best way to export this entire scene and load it in 3DS max?
Another question, if i expand the ability for Bryce 3d to access more RAM do I also need to do this for the dll's?
Daz 3D is part of 
Connect
DAZ Productions, Inc.
224 S 200 W, Suite #250
Salt Lake City, UT 84101
Licensing Agreement | Terms of Service | Privacy Policy | EULA
© 2024 Daz Productions Inc. All Rights Reserved.
Comments
Render To Disc is a tad unreliable... Most of us here only ever use the regular Render button.
You can then export the completed render as a tiff or even a 96bit HDR (other options are also available)
The limitation of Bryce's regular 'Render' is that it will only allow an image of 4000 pixels square, but there are things you can do to end up with an image as big as you want with a bit of forward planning and some maths and some extra work in a Photo Editing application.
It took me a while to find it, but here is an excellent video tutorial from David Brinnen showing how it can be done.
https://www.youtube.com/watch?v=OdZpzQ4t_HQ#t=302
As for exporting your scene to import into 3D Max.... lol, good luck with that. :lol:
Individual elements can be exported in various formats (.obj being the most common), but on the whole, Bryce doesn't play well with other software.
Another thing to bear in mind is to question if you really need to render at 600dpi.
People usually overestimate the resolution they require for finished printwork and a printed dot is not the same as a pixel.
Once you have rendered your image you can usually 'res it up' in a photo editor with no discernible loss of quality.
Hope this helps.
TheSavage64, Your post helped incredibly well! It gave me a lot to think about and study! I will research your instructions and post back here if I run into any snags. Thanks very much for your help!
@RexRed - welcome to the Bryce community. As TheSavage64 said, render to disk is unreliable. As far as memory usage for images is concerned, I have this PDF http://www.horo.ch/docs/mine/pdf/BgMemUsage.pdf that may help you. Bryce, as a 32-bit application can use 2 GB max, if made large address aware, around 3.4 GB.
Forget the DPI in the image, this was invented to confuse. From the size in inch (or cm) the print will be, calculate the number of pixel for the document by multiplying the printer resolution per inch by the size of the final print. Be also aware that a printer with 300 dpi, for instance, has 100 dpi per colour, a 600 dpi one probably 150 per colour R, G, B and black. I made extensive tests a while ago with renders I sent to a photo printer shop. Film industry talks about lines per inch, not dots, but it is essentially the same. I had 112, 140, 179 and 236 lines per inch, the best quality is at 140 lines per inch.
I do recommend that you test render and test print first. Take only a small but representative part of your scene and render it in different sizes and print them at the same size. You will find what is best for the printer you will use. Each printer is different.
Not a better method to render larger than 4000 than David's video TheSavage24 mentioned, just another one. I wrote this tutorial http://www.horo.ch/raytracing/tuts/pdf/minitut21_en.pdf a while ago.
As Horo has pointed out you only have to set the image size to the size of the printed image multiplied by 600.
Going by the sizes you are setting the printed image would be 69 inches by 52.5 inches (approx), at that size a dpi of 150, or less, would do as the image wont be viewed close up, as it is too big. Think of an advertising billboard which is 40 feet by 25 feet, the dpi for that is probably only 4 or 5 dpi but looking from a distance it looks fine. The smaller the image and the closer the viewer is to it the greater the dpi has to be, or you see the dots :)
If your finished image has to be 12inches by 9inches at 600 dpi then the rendered size has to be set at 7200 pixels by 5400 pixels so it prints at the right size.
Something I read years ago which might be interesting.
http://www.scantips.com/no72dpib.html#6
When I save photographs from Photoshop or IrfanView I set the dpi at 1 or 0 and the long side on landscapes at 1000 pixels, unless I am going to print them then I set the size to the printed picture sizex300 (setting it any higher in a home inkjet printer seldom makes any difference as they don't have the dot resolution to cover it. My old Canon one actually printed better at 280 dpi from a 300 ppi image). The 1 or 0 resolution image for viewing on a computer screen is fine as, no matter the resolution of the monitor that it is viewed on, it is still 1000 pixels wide, too wide for an 800 pixel screen and will sit in the centre of a 2000 pixel screen.
@Sandy - When I sent in my renders for printing on photo paper, I rendered the wider side to 1600 pixels which gives 145 dpi on a 11 inch sheet. Now I print them on a 600 dpi printer on 180 gram glossy paper myself and still use this size. I cannot see differences to the screen. Testing and checking - even with a magnifying glass - is the best option to find the sweet size. Quality doesn't degrade if there are more DPIs that can be managed by the printer but the time to render is wasted. And that guy your link points to is absolutely right.
This is from the same site.
http://www.scantips.com/basics3b.html
It explains the printer resolution and why it should be set for the image and not set at 300 dpi all the time.
@Sandy - thank you again. What is repeated? Experiment. You know, this all reminds me of the time when I wrote my text book about CCD astro cameras (ISBN 3-7643-5218-3) and programmed dithering algorithms for the printers in 1994. I spoke for hours with Kodak to get the idea. What this guy writes makes sense, though he starts with a scan, which makes things a bit more complicated because of the interference patterns that can appear.
It is about scanners and scanned pictures but if you skim over that and concentrate on the rest it is easily transferred to photographs and saving for printing. I can't remember when I read it but the copyright says 1997 so it couldn't have been far off then as I was using Amigas before that and then moved to windows 98 when I built my first windows computer. I had already rebuilt the Amiga and had it in a tower with two hard drives and accelerator but the OS fell to far behind for practical purposes so I had to move on. I also don't use the printer to set the properties. I set them in Photoshop and use that to print as the colour correction is easier to set and getting what I see on the monitor is what I get in the printed image. Each image can be different so setting things this way makes sure the image is the best I can get with the products I have.
Thanks people for the invaluable input, It has given me a lot to study. I am still not understanding the camera settings but that is because I have not fully read (and studied) the links you have sent. I have a picture object in my scene and when I link my camera to that it throws my camera angle in the scene off. So I will trying an invisible sphere instead. But setting the FOV, Pan H, Pan V and scale are still quite confusing. I will be very soon reading fully your insightful links and then inquire if I run up against any problems. Again, THANKS! :)
Well, I tried to cheat and I used render to disk. I was having a crash error when I would try and render to disk the scene with anything near 12000 pixels I changed the radial light in the scene to a cylinder spot light and the crash stopped happening. Now when I render anything near 20000 pixels I get an out of memory message before It even starts to "render to disk".
So I was able to render the scene successfully at 12000 x 16000 otherwise it says out of memory. (this render took 24 hours)
My printer person needs the art at 31593 x 42124 (or somewhere thereabouts).
90x120 cm (36 x 47 inch) canvas
I need to know the "magic numbers" on how many squares i need to render the scene and how to set the camera in each render instance.
The math and various parameters of the camera are boggling my head.
Could someone give me some consecutive coordinates for the camera in order to render this large a scene in quadrants?
I am not lazy or anything and I am resigned to figure this out myself but to me this is all new. If there is someone here who could write up the camera coordinates I would need to make a render of this size I would very greatly appreciate this.
Then there is another issue also in play here.
I installed a program called "Large Address Aware". Now it lists Bryce as "yes" being large address aware.
I have 8 gb of RAM and I have been considering bumping that up to 16 gb. If Bryce is only capable, even with Large Address Aware enabled, of seeing roughly 3 GB of RAM wouldn't this ram upgrade be pointless? Or, is Bryce able to utilize all of the system RAM in some instances? Isn't Bryce 64 bit?
My printer person sent a gray scale .tif line drawing at the dimensions of roughly 31593 x 42124. This .tif file was only 13 MB in size.
I had to reduce the image down to 1/4 the resolution before Bryce would accept it as a picture object. Anything larger and Bryce says, "unexpected error generic failure" for a .tif and out of memroy for a .bmp and .jpg. The half size .bmp was only 312mb.
I am worried that once I get the large image rendered in Bryce that the picture object in the center of the scene will be fuzzy looking. I am hoping that 1/4 the resolution for the picture object since it is not filling the entire screen resolution of the overall image will not look fuzzy.
Any comments are greatly appreciated!
Thank you!
(My Graphics card AMD Radeon HD 6850 series 1GB 256-Bit GDDR5)
Going by the finished size he wants it at 900ppi, Why? There is no need for it to be that size at those dimensions as the viewer will be too far away to see the difference between 300 and 900.
This fine art printers only want it at 300dpi so they would only need the image to be 10800x14100 or slightly more for the same print as you require.
http://fineartphotoprinting.co.uk/what-is-a-fine-art-print/
The printer person told me today that DPI was not a critical issue that the resolution was what was important. I am sure they want the art prints to have a nice saturation of DPI even close up because it will be viewed also close because it will not be on a billboard it will be hanging in a gallery where people can come up close and inspect the print also.
I am thinking 300 will probably suffice but I would like to be able to render at 600 just in case. These images were gray scale and when you print black with too few a DPI it begins to look gray and washed out.
I believe it costs 300 dollars just to print one of these items.
I am still in the process of trying to nail down exact details.
The printer person told me today that DPI was not a critical issue that the resolution was what was important. I am sure they want the art prints to have a nice saturation of DPI even close up because it will be viewed also close because it will not be on a billboard it will be hanging in a gallery where people can come up close and inspect the print also.
I am thinking 300 will probably suffice but I would like to be able to render at 600 just in case. These images were gray scale and when you print black with too few a DPI it begins to look gray and washed out.
I believe it costs 300 dollars just to print one of these items.
I am still in the process of trying to nail down exact details. This makes the final Bryce render even more difficult for me to figure out.
I am thinking my out of memory error is caused by my graphics card not having enough memory or both, graphics when loading my picture object and processor when trying to render a large scene..
Does Large Address Aware allow Bryce to access more graphics memory or more processor memory?
I have assumed all along that it was allowing me to use more processor memory, then my question would be would Bryce be able to access 4 GB of Graphics memory if i were to upgrade my graphics card to 4 GB?
I am thinking I will need to render 8 x 9 squares
This would be 72 separate squares i would need to render, I think. :)
Bryce does everything in the processor. It doesn't even know what a graphics card is :)
And no it's not 64bit so about 3Gb is the max you'll get from it. To confound matters more, it also uncompresses everything (no point using a compressed .jpg thinking you're saving memory, because Bryce just eats up the memory uncompressing it for it's own pleasure and your frustration.
WOW, I am glad to hear this, (Incredible tip TheSavage64!) then i will not need to upgrade my graphics card and It seems my ram at 8 GB will also suffice.
I have placed an invisible sphere in my scene and the camera is now tracking it.
Now all I need to do is figure out the overall image resolution while maintaining my 3:4 ratio and figure out the camera coordinates for each render square.
Then start rendering like crazy!!!! (lol)
31593 x 42124 final resolution (I arrived at this resolution because it is very close but slightly over the resolution my printer person said he needed. Also when I divide 42124 by 4 I do not get a long decimal number and when I divide 31593 by 3 I also don't get a long decimal number. But when I divide each of these numbers by 8 I get long decimal numbers. So I have resigned that there will be some overlap in my final rendered squares. I can't seem to resolve this math in whole numbers that are broken down to whole numbered squares.)
4000 x 53333 render resolution for each square
8 x 8 = 64 renders
The patches will overlap but I think I should be able to zoom in and assemble them in 64 bit photoshop as layers.
I am still boggled on how to set the camera coordinates to make these precise 64 renders.
A canvass that is 90x120cm will be a resolution of 10628x14173px at 300dpi.
As has been pointed out, there really is no need for any reason to go beyond that as printers don't use dots anymore.
Originally the DPI was referring to the dot screen used by lithographic printers and is largely redundant in the digital age.
Use halves and quarters etc to work out your tiling, it will make the maths a lot easier and round up if you have to.
So for the pixel dimensions I quoted, divide each side by 4:
you'd need 16 tiles at: 2657x3544 because 4x4=16
If you follow the video tutorial I pointed you to and simply substitute your numbers for the one's David is using (for the offset for Pan H and Pan V and the 100% Camera Scale, you should have no problem.
Assemble all the tiles in a Photo Editor and then when you've got your flat image all stitched together, resample it up to 900dpi if you think it's absolutely necessary.... You'll end up with an image that is about 3.7Gb.... And no doubt that will give you even more headaches.
A canvass that is 90x120cm will be a resolution of 10628x14173px at 300dpi.
As has been pointed out, there really is no need for any reason to go beyond that as printers don't use dots anymore.
Originally the DPI was referring to the dot screen used by lithographic printers and is largely redundant in the digital age.
Use halves and quarters etc to work out your tiling, it will make the maths a lot easier and round up if you have to.
So for the pixel dimensions I quoted, divide each side by 4:
you'd need 16 tiles at: 2657x3544 because 4x4=16
If you follow the video tutorial I pointed you to and simply substitute your numbers for the one's David is using (for the offset for Pan H and Pan V and the 100% Camera Scale, you should have no problem.
Assemble all the tiles in a Photo Editor and then when you've got your flat image all stitched together, resample it up to 900dpi if you think it's absolutely necessary.... You'll end up with an image that is about 3.7Gb.... And no doubt that will give you even more headaches.
Thanks again for this excellent help. In my previous post I am not disagreeing with you. I am just trying to make sense of this.
And I am still swimming in numbers.
I got an email from this printer person today who said
"Sorry but,
First 12000x16000 pixels is very too small for a large print of 90x120 cm (36 x 47 inch)
Half this dimension, perhaps ...
Regardless of the number of dpi only the pixel size is important.
Second, if you are limited in pixel, I must deliver another file.
It is certainly not perfect to interpolate from a large file because the width of the line must be changed."
Comment:
It is difficult to understand with his own translation to English that he sent in my email.
It seems I am getting conflicting Information here.
He wants art in the 30000 x 40000 range.
It seems he is saying that 12000 x 16000 will only print half that size of 90x120 cm (36 x 47 inch)
I think he is saying the fine lines in the drawings become enlarged when a smaller piece is scaled to this large size.
But I may be interpreting this wrong but then why did he send me his print masters that were 31592 x 41454? His print masters take 20 to 30 seconds to open in windows picture viewer.
He also sent me another master image for a logo. I took his file and simply added text to it and then saved it as is and sent it to him and this is why (when he says "secondly") he is sending me a larger file because he says even this was too small (about half) for the large prints and that he made a mistake and he is sending me the logo image masters probably in the 30000 pixel range. (They want to print and market the Logo art I made huge even though it will also for the most part be used as tiny web graphics.)
I am making two pieces of art for them, one I am simply adding text to and the other, a different design, I have inserted into Bryce.
I have rounded my numbers to a resolution of 30000 x 40000
I made this graph
When I enter in any of these coordinates to the Pan V and Pan H boxes it seems the render is zoomed in so far that there is no object or detail in any of the renders just solid color. I set the document setup to 4000 x 5333 and also 3000 x 4000.
What am i doing wrong?
Rex,
From what I've read so far from this thread you will certainly be forced to render the image in tiles. It seems that Render to Disc is slower than rendering in the normal viewport so try not to render to disc for the tiles.
Best of luck.
Not sure where you are but I did an example on this site http://fineartphotoprinting.co.uk/what-is-a-fine-art-print/ for the size you require and the quote came up at £86.96 including vat. They only require the image to be 300dpi at the pixel resolution for the finished print which is 10800x14400pixels.
Which is 86.96 GBP=139.392 USD At today's rate.
Screen shot below.
Not sure where you are but I did an example on this site http://fineartphotoprinting.co.uk/what-is-a-fine-art-print/ for the size you require and the quote came up at £86.96 including vat. They only require the image to be 300dpi at the pixel resolution for the finished print which is 10800x14400pixels.
Which is 86.96 GBP=139.392 USD At today's rate.
Screen shot below.
The hidden cost of these prints may be the special canvases that are crafted in Holland, It may also be that there is a lot of ink that is on the canvas versus a light spattering. It may be because these prints require a very deep dark pitch black that is used. I am not certain and I am still totally stumped as to why the extremely large resolutions are needed. I have not had to personally purchase these prints so I am not really sure of the canvas prices, processes and the printing methods.
I have made the matrix above out of the odd possibility that I would need to render a very large image in pieces.
I would like to agree with TheSavbage64 because if his calculations are correct, I have already finished the second piece of art needed that I rendered in Bryce. Yes I used render to disk, yes it took 24 hours to render. The dimension are 12000 x 16000.
According to his calculations that is well within the range needed to be printed on a 90x120cm canvas.
I am not at all knowledgeable enough to contradict or doubt his very useful advice.
I am just totally stumped as to why this printer person in France needs that resolution doubled.
Could he be printing on some other type of proprietary surface that requires more dots? I don't know. I have seen some of his prints and they are loaded with ink and have a very pungent smell! I pulled one of the prints out of the protective plastic liner and was nearly knocked out by the fumes and the black was so dark that it looked like a deep black hole. I don't think I have ever seen anything in my life that black.
I don't want to contradict him and and I don't want to contradict TheSavage64 but at some point I am going to have to remark that I have been told that this 12000 x 16000 resolution Bryce render should print fine on a 9x120cm canvas.
But just in case I need to in the end supply him with this ginormous resolution image that he is demanding I want to have my homework done.
As for David's video... I have to say it is outwardly helpful but it has not led to a fruitful outcome for me.
When someone says here is how it is done to print in sections a very large image in Bryce (but let me scale it down to 400 x 400).
Why did he not render the large image and just chop out the rendering in the video?
How do you set the scene for very large renders? He set it for small renders, how would you set the document setting for a very large render? That part is not "clearly" discussed in the video.
In trying to render this large image I have had Bryce several times crash. And this is not by rendering to disk but using the regular render at the normal antialiasing setting.
I only have this very large picture object, a cylinder light, a plane and a sphere in the scene with atmosphere on.
I have decide the size of my large scene i want to render, I have rounded the large numbers to whole numbers that are easier to cut into fractions.
30000 x 40000
I have broken this up into 64 quadrants in the matrix image I uploaded.
I would think I simply add the quadrants into the pan left and pan right boxes.
But how do I set the document resolution for a very large scene?
And why did David set his camera scale to 200%? One would think if you scale a camera view port to 200% it would widen rather than narrow?
There is something I am missing here.
I am really trying to figure this out.
Let's say that TheSavage64 is right but now I need to print on a canvas that is quadruple the size and the printing needs to be fine for close up views also.
How can one say this is how you print large but then scale it down for a tutorial?
David set his document setup for 400 x 400? That does not show me how to set it for very large renderings?
Do I need to scale the camera down to 1000% is that the parameter I am missing here?
There is something I am missing in this tutorial and I have not figured it out yet.
Why did David scale his camera?
The math does not seem to add up (to me). If you have a 400 x 400 and you have 4 200 x 200 tiles why not leave the camera view at 100%?
Perhaps it is by scaling the camera view smaller this makes the image larger like shrinking a person makes the world larger?
This does not on the surface seem to be covered in the video.
Something here is eluding me. (lol)
@RexRed - the TIFF file (post #10) is only 13 MB at the size of 31593 x 42124 because it is monochrome. Bryce automatically makes a colour image of it, even though the R/G/B values are the same and therefore Bryce uses 40 MB of memory.
LAA (large address aware) gives at best 3.5 GB for Bryce. Bryce is a 32-bit application and all it can theoretically address with the LAA flag set are 4 GB, without it 2 GB. If your machine already has 8 GB, upgrading to 16 GB won't help Bryce at all.
Bryce does not use Graphics memory except for display (as TheSavage64 already mentioned). In fact, I could never see a difference between a lousy and an expensive graphics card as far as Bryce performance is concerned.
Bryce is importing the tiff as a picture object but I have loaded the image twice so it will double whatever it increases each file to. I loaded the image twice so the inverse would turn the black part of the line drawings transparent leaving only the white lines. Then I shone colored light on the white lines to change their hue.
Before I even loaded the second image and while trying to load the first image Bryce complained it was was out of memory.
The image disk size is small but the resolution to begin with is so excessively large.
I tried converting the image to jpg and bmp before importing into Bryce and that did not change Bryce's memory complaints.
Perhaps it is the sheer resolution that simply brings up an out of memory error, though it may not be out of memory. Though the size is quite excessive but is it excessive enough to swallow up 3+ GB of ram? I can't imagine that Windows itself is using much more than 1 GB.
I also did a ctrl, alt, delete, and shut down any background process that might have been using memory, to no avail.
What exactly is a picture object comprised of?
This conversion from an image into this picture object must exponentially use up RAM. It seems it is doing more than simply converting it to a bitmap where maybe the pixels of the image become a Bryce "texture"... It is certainly a quandary...
@RexRed - Bryce blows-up any image loaded to a full bitmap. Loading a compressed picture like a JPG is actually worse because it loads the JPG, decompresses it and creates the bitmap. The memory used for the JPG is later released but for a while, the memory is used.
There is the free tool Process Monitor I mention in my video Bryce memory shortage and what you can do about it. It can show the memory used by an individual program. The Task Manager is not helpful because it shows the total memory used. Try this little tool, Start it, start Bryce and then find Bryce in the list of the Process Monitor. You can see directly what happens with the memory.
If you have two screens, that would be cool because Bryce covers the main display. If you've got only one, minimise Bryce when you load a picture so you can see what's going on. This will help you to figure out whether there are memory issues and if, at what point they present themselves.
I have another question also.
pan v and pan h
It is apparently panning the scene while the camera is remaining stationary.
Is it panning the scene relative to the camera's position or relative to zero XY and Z?
Why did David place the sphere object in the scene and link the camera to it. Does this serve any purpose in how pan v and pan h works?
And while I am at it, I noticed David has moved his camera to the bottom of the scene and pointed it up (+ Z I think). I have always done this.
Why does the default scene in Bryce set that camera at a 45 degree angle in the top right hand quadrant looking towards zero.
At that angle everything is at a 45 degree angle and then it is hard to figure out the fractions when moving things around?
Is pan h and pan v moving the scene from that default orientation or from the current camera location?
Would it not make more sense to orient the default camera at the bottom of the scene +Z looking towards zero than at an angle in the top left quadrant?
What if i wanted to have my camera look towards the sky would the pan h and pan v move the scene in relation to where the camera is looking or move the scene in relation to where zero is?
Why did David place the sphere in the scene for the camera to track and why did he re-size it then land it to the ground?
Why did he not place the center of the sphere at zero?
The picture object in my scene is a bit beyond zero a bit and my camera is looking there beyond zero and not directly at zero.
So my invisible sphere (which I tracked my camera to though I do not know why this is needed for large rendering) is beyond zero further into my scene so my camera points at my picture object. at an angle sloping down a bit.
Redundant reply not sure how to delete a reply yet.
OK, just to potentially confuse matters even more.... I'm just checking here if you are setting up your document correctly and then using the scale render settings in the document set up window?
What I show below is that to get a finished render size of 3000x4000px per tile, your document should only be set up to 750x1000 at 1:1
You then render it at 1:4 (highlighted in red on my picture).