Everything to do with Lighting in Bryce 7.1
This discussion has been closed.
Daz 3D is part of 
Connect
DAZ Productions, Inc.
7533 S Center View Ct #4664
West Jordan, UT 84084
Licensing Agreement | Terms of Service | Privacy Policy | EULA
© 2024 Daz Productions Inc. All Rights Reserved.
Comments
I think we can overlook the occasional misspellings *gasp David is human* so long as you keep making the discoveries you're making. :)
Thanks for the tutorial, it's just been moved into the "next slot" on my tutorial que. Hopefully with it's help I will create a convincingly realistic prescription label with which I can practice creative license on :roll:
Well after reading the lighting discussion of the last page I can't help but feel as if I'm active in a thread that may one day have historic significance in the Bryce realm. I don't understand most of it but hey at least I'm here, surround by smart people....all starting to glare at me for distracting the conversati....okay I'll shut up now. :)
Mark,
I am certainly not the one staring. Peter, Horo, and David are truly smart individuals. I am just an extremely determined "pretender." Beneath all the artsy fartsy is solid mathematics which one either has the background to keep with up or they don't. I fall someplace in the middle where I get most of the math, but not all of it like Horo or Peter or even David. Ask me to write an equation to calculate the surface luminance of an object and I'm sure I could probably do a good deal of it, but I'd probably not have the chops to complete it. Horo on the other hand, would. For some reason I seem to have a knack for knowing when light appears to behave naturally in a render, and an ability to recognize even if I cant describe it concisely, when the light doesn't behave naturally. This lead to much discussion (some of it pleasurably nasty) regarding TA and other tools back in the days of Bryce 6 and earlier. Now that some of us have had the benefit of discussion with actual programmers who can see the code and algorithms we can have more productive discussions than we used to. Hence, the last few pages.
In theory TA especially with Boost Light enabled is the most technically tricky lighting tool to discuss once you have gotten the hang of IBL concepts and the idea of high dynamic range light and color. In the case of lighting threads I tend to initiate, I am all about taking the guess work out of it all. I want solid consistent solutions that are easy to use and comprehend, solutions that simply work, solutions that are as accurate as possible. For those solutions that dont yet work, I am very much interested in finding out WHy they dont work so i can figure out a way to make them work down the line when I need them. I am not alone in this approach, I am just as stated above an extremely determined pretender of sorts.
This thread will be significant. Even if Daz is not developing Bryce currently, Horo, David, myself, and several others are doing what we can to continually develop Bryce. Many of the tricks we discussed in earlier lighting threads matured into full features in Bryce 7 we all can use now such as Domes, 3D Fills, TA Optimization and much more. It would not surprise me in the slightest if this thread also yielded useful tricks that could be turned later into mature features for Bryce 8 or later.
All that to say, you are very much welcomed to this thread. A thread dominated by a bunch of bitchy old biddies (no intended offense to anyone) discussing the most subtle of subtleties is not the sole point of this thread, though some degree of that is expected and encouraged in this thread. So stick around and please feel free to distract the conversation as you see fit.
Not as good, probably not even useable – but certainly possible. This Renderosity thread from nine years ago (!) features me pratting about with caustics using Bryce5 TA.
http://www.renderosity.com/mod/forumpro/showthread.php?message_id=1445509&ebot;_calc_page#message_1445509
Of course, David B. working with today’s improved TA options is way, way ahead of my earlier fiddlings. But they do show the potential was there. “If only there could be more light,” I said at one point. Well now there is, and David (B for brilliant) has found a way of reducing its horrible side effect.
You’ve all done very well…
.
.
Thank you for the link. 2003... before internet here I think. But I did have Bryce 5 from late February onwards. And I did tinker with TA and thought "Hmn... this seems like a good way to bring my CPU to boiling point for no very good reason". I'd just upgraded to my AMD 750 Mhz system and was playing Planescape Torment... Sorry I'm getting lost in memory lane.
So to your thread....
Talk about re-inventing the wheel! All very familiar territory. Looking at the renders I am suspecting that the caustic effect created by the light travelling through the glass was a result more of the less scattered nature of legacy TA than correct optics - however I will re-examine my tests with lenses to see if I have missed a trick.
It is clear, you have nicely predicted what was possible by a cool 10 years!
Please, if you would be so kind, email me with the numbers for next weeks winning lottery ticket (preferably before the ticket is drawn).
Mark, well done getting the models into Bryce and that's a nice looking render - it must have taken a while?
What I would suggest, rather than passing these bulky models around, is if you write up a tutorial on how to get them, convert them and import them - post it up on a thread here and I'll get my friend to transfer it over to Bryce-tutorials and then it will be there as a reference for other Brycers to take advantage of.
What do you reckon?
Speed is only part of the consideration. Accuracy for me is also important as I perceive accuracy as telling more of the truth rather than hiding the truth with lies, which fudging equates to. Still, I know well that most of us CANT HANDLE THE TRUTH. So a little fudging probably is in order at all times.
I am looking for a way to increase both accuracy and speed, while also lifting Bryce rendering into the new century.
Octane Render is currently the fastest unbiased render that I have heard of and it is GPU based.
http://render.otoy.com/
Clearly, there is no point in reinventing the wheel. No matter what we do with TA, it will take a bit of work to get to the level of Octane Render. What I think we need is to do a bit of what DazStudio has done, and cozy up to one of these unbiased render solutions. I would actually prefer Octane to Luxrender, because super fast rendering is something Bryce has not been able to boast of in nearly two decades. This GPU stuff seems promising to me, makes CPU discussions almost seem irrelevant in terms of where to go in the future. GPU all the way.
Am I wishing upon too distant of a star?
Thank you for the link. 2003... before internet here I think. But I did have Bryce 5 from late February onwards. And I did tinker with TA and thought "Hmn... this seems like a good way to bring my CPU to boiling point for no very good reason". I'd just upgraded to my AMD 750 Mhz system and was playing Planescape Torment... Sorry I'm getting lost in memory lane.
So to your thread....
Talk about re-inventing the wheel! All very familiar territory. Looking at the renders I am suspecting that the caustic effect created by the light travelling through the glass was a result more of the less scattered nature of legacy TA than correct optics - however I will re-examine my tests with lenses to see if I have missed a trick.
It is clear, you have nicely predicted what was possible by a cool 10 years!
Please, if you would be so kind, email me with the numbers for next weeks winning lottery ticket (preferably before the ticket is drawn).
Peter is an incredibly fascinating guy. That thread was a bit like reading a prophecy. Scary almost! I feel it necessary to try that mirror test again with Boost Light and TA Optimized light sources. So glad I have the day off today!
Oh I was only joking about distracting the thread, I just said that because I know how serious people can get when discussing the technical aspects of things they are passionate about. As for you being a pretender well keep in mind, that how you see yourself compared to the likes of David and Horo but that's proabably the only person that sees you that way. Me personally I see you as a very dedicated and enthusiastic member of the steering committee. I'm not sure if this is your intended job description in the committee but what I see you doing is keeping after people to keep improving things rather then get too complacent.
As for the historic comment, well I can easily see similar discussions having happened in the past that led to changes in Bryce 7 so likewise I could see this thread having implications for Bryce 8 and beyond. Oh and on the issue of math. Let me explain it this way. I was fine with math al the way up until high school. Then I had to take either Algebra, Calculus or Geometry. I had always heard the first two were very hard so I passed on them and went for Geometry. Which was fine, I always liked geometry before because it mostly involved drawing shapes and using compasses and protractors. In high school though it was nothing like that. Instead it was mostly memorizing a ton of proofs and theorems which bored the hell out of my hormone enhanced teenaged mind at the time. Needless to say I didn't do well in math.
Well there isn't much to write up. All I did was install meshlab (free app available here: meshlab.sourceforge.net) open it up, select the option to "import model" which will allow you to select any supported format which in this case was .ply. Then after that I selected, "export model" which lets you export it in a number of formats including .obj. Once it was in .obj form I then imported it into studio, scaled it if necessary (some of the items come in like 10 times bigger then your viewport) and then sent it via the bridge to bryce. The trick was just waiting long enough for it to happen but for you guys with your muscle machines for computers shouldn't need as much patience as I had to muster.
As for the time it took to do the render, well I made it take a long time even though I didn't have to, just to have some cool looking materials on them. The solid diamond brain and the ruby golfball is what hurt the most though. Those two alone tripled the render time. I also used soft shadows to use the premium render and set the rays to 256 per pixel for the final draft as you suggested in the capsules video but if I had left it there I'd probably just be posting the picture now instead of last night. Instead I cut it back to 144 rays per pixel which reduced the render time from nearly 24 hours to around 14 hours
Well there isn't much to write up. All I did was install meshlab (free app available here: meshlab.sourceforge.net) open it up, select the option to "import model" which will allow you to select any supported format which in this case was .ply. Then after that I selected, "export model" which lets you export it in a number of formats including .obj. Once it was in .obj form I then imported it into studio, scaled it if necessary (some of the items come in like 10 times bigger then your viewport) and then sent it via the bridge to bryce. The trick was just waiting long enough for it to happen but for you guys with you muscle machines for computers shouldn't need as much patience as I had to muster
Thanks Mark, well, I'll give it a go and see how I get on.
In the mean time, a compare and contrast.
1 - a scene made in 2007 and rendered in Bryce 6 using many conventional radial lights.
2 - the same scene updated today, rendered with three TA optimised gel lights driven by a spherically mapped ( Spherical Mapper ) capture of the HDRI backdrop and one blocking gel to isolate the TA rendering from the IBL backdrop which drives the reflections. Bryce sun provides specular only to drive material specular response. In essence the same approach as taken in this video Bryce 20 minute scene lighting project - Using IBL with boost light and TA gels - by David Brinnen
Most notable about the new scene, better contrast and colour reproduction, no shadow bands to contend with. Render time 1 hour thirty - about 10x longer than the original.
Well I have now successfully managed to get the two statues that wouldn't import to finally import. The problem was the sizes because in both cases after some serious decimation of the mesh I did get them down to managable sizes. the one that goes with the xyzrgb dragon had one billion faces originally. I had to get that number down to closer to one million before I could get it into Bryce in a reasonable time frame (about one minute). The angel statue named lucy I neglected to get the before and after number on the faces but suffice it to say I had to reduce it to about 25% of the original. I'm rendering pictures now to show you all later. I suppose I could post up directions on how I reduced the mesh although it would be a bit involved. I must say from a novice perspective that Meshlab seems like a pretty versatile tool for modelers.
Thanks Mark, I'm going to get that Mesh Lab installed (haven't found it before) and get those Stanford thingies again. We'll see whether I can manage.
Mark,
Those pills look fantastic! Very realistic. The only quibble I can generate is that the division between the green and purple areas is too soft, it should be much sharper. I think the impression should be of a capsule made of two pieces one fitting neatly within the other. Otherwise, it looks perfect. Great work!
Hello, I'm quite a new on this forum. I'm self thought graphic designer, a bit of programmer, CG admirer and occassional beta tester of some apps.
I hope you don't mind me hijacking this thread while my travel to look for true refractive caustics in Bryce ;)
I started independently testing Bryce renders engine capabilities and I have some observations.
Renderer in Bryce looks for me like a brute force forward path tracer.
- http://web4.cs.ucl.ac.uk/teaching/4074/archive/2010/Slides/JK2010/03_path tracing.pdf (what PT is - easy Path Tracing maths course)
In classic path tracing the ray is shot from the specific pixel and then it travels bouncing at random direction from the diffuse surfaces, at reflection angle from mirror surfaces or is bend when going into the refractive objects, and it bounces till it finds light source. Let's consider two situations:
1. exterior environment with open sky
2. interior with small light source
In 1st situation the sky is being found quite quickly terminating the further ray bounces and rendering becomes quite fast. In situation 2 the rays are bouncing till they find a light source or reach maximum depth. If the light source is very small there's no guarantee that bouncing ray will ever find the light source. What more in Bryce the TA optimized lights are probably regular meshes (primitives) with the direct light inside them that is including for taking account into lighting only the light mesh. I've made a model of such light in Bryce 5 by using sphere light inside the sphere with 'range' so that only the sphere is lit. It emits the light of maximum strength of 1. Scene has TA enabled (in 5 it's legacy one) and there's no Boost light switch. Image 1b.png
This way of rendering area lights give a greater way of control for potential user, but at the same time is wasting time for rendering single invisible diffuse ray bounce. True area lights with importance sampling would be a single bounce quicker and would had better distributed samples.
More light was allowed in 7 with Boost light option and as shown by David Brinnen in his tutorials is often created by amplifying the light on inside of the sphere or cuboid with partially transparent 'shell'. I've read here http://www.kerkythea.net/joomla/index.php?option=com_easyfaq&task=cat&catid=19&Itemid=41#faq25 that creating physically correct, non amplifying light materials is a helpful way of getting rid of fireflies. So I altered the characteristics of TA optimized light material (I used fuzzy, 25% transparent, 75% diffuse, 75% ambient) and it changed a little bit light directional response and totally got rid of fireflies! (You can compare 1st image from 4_mix.png with 2 image from 5_mix.png - I'll have to rerender 1st from 4_mix.png with new material to confirm my observations).
Another ways of getting rid of fireflies in Path Tracing in this article: http://wiki.blender.org/index.php/Doc:2.6/Manual/Render/Cycles/Reducing_Noise Especially 2 options are interesting, so I'd cover them briefly here:
1. Sample as lamp - is directly connected to what I wrote above. It means that TA optimized light is possibly sampled like a regular mesh instead of the light source (well... I'll have to investigate it a little bit more, and try out TA optimized domes and fills)
2. Clamp fireflies - is option inverse to boost light, but instead of limiting light to 0.0 - 1.0 range it allows setting the upper range to any value. Setting clamping at value higher than 1 results in getting rid of fireflies while making light response natural http://archive.dingto.org/2012/blender/clamp_overview_bmw_mike_pan.jpg
My feature request for future Bryce versions would be an option:
Clamp high (checkbox and input field - values from 0 - +inf) - if the light strength is higher than the value use value instead - option for reducing fireflies
Discard low (checkbox and input field - values from -inf - 1) - if the light strength is lower than the value then stop tracing further rays - option for stopping tracing rays and speeding up rendering or getting rid of black fireflies when max tracing depth or clamp when using negative lights
Several posts ago somebody mentioned the alternative ways of rendering Photon mapping, but I think it's not very 'mega scene'/instancing/displacement friendly. Storing big photon map makes use of memory and often is not good idea for big scenes. It however works well for interiors and especially for caustics, it would be nice addition for rendering caustics but imho not a primary method (comparison of rendering methods: http://raytracey.blogspot.com/2011/01/which-algorithm-is-best-choice-for-real.html )
Back to the refractive caustics... I think I've managed to catch the real ones :D (1st image from 5_mix.png compared to the second rendered with similar setup using direct light)
Now I need to try to render them in the mirror ( http://www.indigorenderer.com/forum/viewtopic.php?f=37&t=10142&start=15 ) to see if they'll disappear - if yes - then they're 'real'
BTW.
There might be a bug of how TA treats the face turned to the front compared with the face turned to the right. Please have a look at the colourization on the back face coming from the red face compared with the blue one, it might not be caustic pattern.
Regards
dwsel
Thank you dwsel_ for joining in the lighting thread with your insights. The fill lights in TA optimised mode are interesting, the TA light inside is generated from the hulls of the many individual tiny light sources bounded by the total shape of the dome and fills. They render a bit slower then radials in TA but seem to offer (I've not got very far with testing these) a sort of volume TA solution. I don't know what you'd call this, it would be as if all the space was glowing and emitting light in a great invisible cloud of light.
Well if you look at David's tutorial the capsules are actually off of a primitve you contributed to the additional primitives section of the object library that comes with Bryce 7 Pro. He then shows you how to apply a material from a scene from the bonus content and then split the material in two across the object to create the two sides of the capsule. At least that's my interpretation of what he showed me in the tutorial. Now you are correct in that it does lack that slight bit of realism of one half being ever so slightly larger then the other to allow the two halves to fit together, creating a subtle edge that can cast a slight shadow. I'm not sure if there is a particular reason why the tutorial does it the way it did rather then making an actual capsule shape from two objects? I would offer the guess that doing it thus would require less exploration of the abilities of the material lab thereby making the tutorial less informative?
I think when I get the label done and people viewing the image start reading that, they'll be less likely to notice the lack of an edge around the center.
Well if you look at David's tutorial the capsules are actually off of a primitve you contributed to the additional primitives section of the object library that comes with Bryce 7 Pro. He then shows you how to apply a material from a scene from the bonus content and then split the material in two across the object to create the two sides of the capsule. At least that's my interpretation of what he showed me in the tutorial. Now you are correct in that it does lack that slight bit of realism of one half being ever so slightly larger then the other to allow the two halves to fit together, creating a subtle edge that can cast a slight shadow. I'm not sure if there is a particular reason why the tutorial does it the way it did rather then making an actual capsule shape from two objects? I would offer the guess that doing it thus would require less exploration of the abilities of the material lab thereby making the tutorial less informative?
I think when I get the label done and people viewing the image start reading that, they'll be less likely to notice the lack of an edge around the center.
You are correct Mark, I tried the tutorial both ways with using the capsule and fiddling around with boolean operations and making one side slightly larger than the other and all that, and once I was done I thought, yes this is all very well, but it is just manipulating objects and for a tutorial it is time consuming watching me faff around and it does not really teach anything that isn't already known. Other than I am slow when it comes to boolean modelling. So I decided to incorporate a bit of material manipulation in as well, because it could be done faster and it offered a bit more encouragement for folks to play with the material lab. That was my thinking.
You're my hero! I've installed MatLab, got the Stanford models and converted them to obj. I imported them directly into Bryce - why use Studio and the bridge? Here are my results:
Amadillo, 26.7 MB as obj, uses 73.5 MB of Bryce memory and imports in about 5 seconds.
Bunny, 6 MB as obj, uses 24.5 MB of Bryce memory, imports in about 1 second.
Dragon, 68.4 MB as obj, uses 168 MB of Bryce memory, imports in 30 seconds.
Happy Buddha, 85.5 MB as obj, uses 185 MB of Bryce memory, imports in 1 minute.
xyzrgb_Dragon, 615.4 MB as obj, uses 1.3 GB of Bryce memory, imports in 39 minutes.
Now, the file could not be saved, neither could I put the dragon into the Objects Library, each time I got an Out of Memory error, even though there were still 2 GB left. But Bryce did not crash.
So I exported the object as 3ds, which took about 5 seconds. Adter quitting and restarting Bryce, I imported the 3ds with a filesize of 150.4 MB. 1.3 GB of Bryce memory were used and the dragon imported in around 10 seconds. I could save it in the Objects Library which grew by 618.2 MB.
The dragon can be taken out of the Object Library, which takes about 10 seconds. Bryce memory usage is 1.3 GB. I've monitored memory usage of Bryce only with the Process Monitor.
I still have to import xyzegb_statue at 864.1 MB as object. I reckon that would take about 2 hours. And then, there is Lucy at 2.5 GB as object - that one will probably take a day, if it imports at the full resolution.
Thanks again Mark. Without your comment, I wouldn't have tried again.
Er, no, my head is a maths free zone. No idea where you got that notion from. If it was from me then it was just a calculated lie.
Perhaps you meant moths - I have any number of those sodding things. Probably the only way to subtract them is to burn the house down. I'm weighing the pluses and minuses on that...
.
Dwsel,
Fascinating!
That's a very clever insight I certainly hadn't considered. Storing maps would consume much of the limited resources Bryce already has. In my mind I would assume the exchange of probe discoveries among pixels would be no more challenging to calculate than mesh smoothing, but I guess its a bit trickier.
Regards
dwsel
There was a bug with the default cube from Bryce 6 and earlier. It was repaired in Bryce 7. If you are recreating scenes from your Bryce 5 days it's possible you've imported an old cube. Not sure if would really matter since primitives aren't real geometry anyhow. Just stabbing in the dark on that one.
Cool glad to be of help however, as happy as I am for you to hear your results it begs the question of why my Bryce 7 Pro installation can't import .obj's at all. I was assuming it was something odd about the models as a result of the conversion but you used the same program and it worked. There is an option in the filters section of meshlab that will let you reduce the mesh significantly which might make things easier for you? I can attempt to work up the steps I followed should you have need. That's how I was able to get Lucy and the other xyzrgb statue over the bridge and into Bryce. Here are some renders to show you how they look.
Obviously these images are based on the greatly reduced meshes since that's the only way I could get them in Bryce and at least as far as my eyes can tell I see no negative side effects of that mesh reduction. Although I did use the least reduced mesh I could get away with. I do know some of the even more reduced versions you can see surface differences on the mesh.
There was a bug with the default cube from Bryce 6 and earlier. It was repaired in Bryce 7. If you are recreating scenes from your Bryce 5 days it's possible you've imported an old cube. Not sure if would really matter since primitives aren't real geometry anyhow. Just stabbing in the dark on that one.
The weird thing is that the scene has been built in Bryce 7 from 2D faces and square parallel light only. I guess I'll need to render more cornell boxes to investigate that.
Anyway...
Here are my new caustics tests. I've got the feeling that my patience along my computer's speed wouldn't handle rendering a lot of these on daily basis ;P
These objects are from default library (build from booleans). The water is not correctly set up yet for this test (there's a gap between the water and glass which should be in reality substituted with glass-water interface of different IOR). The look of the caustics depend highly on use of TIR - it looks that it should be on with risen number of bounces, but at the same time it looks that TIR 'steals' rays from regular ray depth, making rendering faster but causing glass to be black or incorrectly distorted in some places. This image uses Ray Depth of 17 and TIR Depth of 5. I hope it was not overkill.
On image there's comparison with default fake caustics for direct light.
@LordHardDriven - The xyzrgb_statue has 864.1 MB as obj file. Bryce needs 1.81 GB and 70 minutes to load. Unfortunately, I could not save it, neither convert it to 3ds.
I haven't yet figured out how to reduce the meshes. Obviously, a 600 MB obj file is about all Bryce can handle even when made large address aware.
The Wavefront obj file has been with Bryce since at least v4. I could import the bunny as obj into Bryce 4 within seconds. I can't imagine what's wrong with your Bryce 7.1 installation.
@dwsel_ - your comments about different render methods are very interesting. Thank you for sharing them with us.
Well it's not a time thing with my installation of Bryce on importing .obj's. I get an error message instantly, that reads, "An unexpected error has occured (failed creation)." The slow time on imports were only happening to me on sending the same obj's over the bridge from Studio. The bigger the file size of the mesh the longer it took.
As for shrinking the mesh well here is how I did it. First I converted the files of course and then once I had them in .obj format I then imported them back into meshlab using the .obj I just created. Once it was fully loaded I then went to the filters menu of meshlab and from the drop down menu that appears I went to the catagory of selection and then on the secondary drop down menu for the selections catagory I chose the option select all. A box then popped up and in that box I clicked the option to apply. You should notice the model turn red from the selection process (a deep red because the mesh is so dense on some of these). Then I went back to the filters menu and this time I went down to the Remeshing, Simplification and Reconstruction catagory and from the secondary drop down menu I chose the option "Quadric Edge Collapse Descimation". A box pops up that shows the default setting for the process and also allows you to put in your own settings if you prefer. The default setting typically cuts the number of faces in half and I know if you push this sort of process too far it'll seriously degrade the mesh. So I just left it at the default settings and clicked apply. You'll see a progress bar at the bottom showing you how the process is going and at the end you'll notice the number of faces reported below the model will have been reduced to what was set in the default setting of the decimation process. Now since I wanted to preserve the quality of the mesh as best I could I then saved it closed meshlab to clear anything out of memory. Opened it again, imported the reduced mesh and then repeated the process. I did this 4 times so as to have several to try and I'd go with the first one that I could actually get in Bryce. That way I'd have the least reduced mesh that worked and therefore the mesh quality should be as good as it will be. Which as near as I can tell the mesh quality wasn't severely impacted.
Okay I finally had a chance to do the label tutorial and below is the result of that. There was an issue though that I imagine was the result of how the lighting was achieved. Besically the colors in the texture for the label got washed out. The little pharmacy icon in the top left corner should be orange not grey. Also the store name of Hulk-Mart was green and there are also some yellow boxes behind the prescription number and where it says no refills Dr. must authorize, that don't show up. Ive also included the texture I used for the label so you can see what it was supposed to look like.
Another odd thing was that I did things exactly as described in the tutorial for the label with the only difference being that I used photshop instead of paintshop. In the material editor everything looked fine as I set it up but the text on the label wasn't showing up. Now in the past when I've used masks I've always had them on a seperate channel from the texture. Now also in the past when I've two things on one channel it's been with bump maps. Instead of using a seperate bump map though I've always just used the texture as the bump map in the texture source editor by copying the texture on the left and pasting it in the spot in the middle to get the desired result on the right. So I tried that for my label and it worked and the text showed up but it shouldn't be that way should it?
@LordHardDriven - I can't help you with the colours. I've never had this issue. In what format have you saved the picture out of Photoshop?
Been experimenting with MeshLab, crashed on me several times. I had a hard time removing that "XYZ = RGB" below the xyzrgb_dragon but finally was successful.
Your help to reduce the meshes is very much appreciated. I halved and saved as PLY, which is faster. Then halved this one as well and had a crash. Had to re-launch MeshLab. I did all manipulations in PLY. It was not possible to reduce the mesh count from the original 4 times. Oh, it could be done, just on saving it crashed. I finally relaunched it after each PLY export, halved, saved as PLY, quit, launched, ad infinitum (well, not quite). Then I loaded the reduced PLYs and exported as OBJ. I had the impression things run faster this way. I lost the whole evening on the xyzrgb_dragon but could test the xyzrgb_statue at 1/8 resolution in Bryce and it looked still quite good.
You must understand these modelling thingies quite well. I had looked at the menus yesterday and was just baffled. In fact, I still am but with your help, I could accomplish what I wanted.
I think I know exactly what has happened here, post me the file and I will fix it for you - and in the process I will explain. It is easier if I show you with your own file.
Actually no, I don't really understand them that well, just enough to make me dangerous :) Plus like I said, the version I have anyway gives you a description of what the options do when you move your mouse over the item, so that helped guide me a bit. Basically I just started looking for words that might relate to reducing a mesh. I saw the word decimate and zeroed in on that because Daz has a program for reducing meshes called decimator.
The way you did it by working from the .ply format was probably the smartest way to do it, reduce the mesh in it's original form first and then convert it.