Page 41 of 47
Re: CoolColJ's test pics thread
Posted: Sat Jun 12, 2010 12:39 am
by CoolColJ
Still getting lots of random BSODs even when doing nothing.
I dropped back to 1x6Gb kit and it works fine, then I try to run this kit at 1500+ mhtz, it's rated for 1600mhtz it crashes... swap to the other kit and try and run it, it runs fine at over 1500mhz....
Ok it looks like 1 kit has a damaged stick in it, but it tests fine in the windows memory tester...
Guess I return it and try and swap for another set, if not get my money back for that one
edit - With the good kit, I can boot fine into windows at 4.2ghz, with pretty conservative voltages and at the full 1600mhtz ram speed. Damn I should have done this last week, would have saved me much headache!!!!
reran the test scenes again now at 4.2ghz and full ram speed in v2.4.3 - better numbers
caterpilla 740k/sec
Erotica 1126k
Pins 862k
Re: CoolColJ's test pics thread
Posted: Sat Jun 12, 2010 5:32 am
by dag
You're running with 2T command rate?
What's the white stuff on top of the tree in second render?
Re: CoolColJ's test pics thread
Posted: Sat Jun 12, 2010 12:04 pm
by CoolColJ
dag wrote:You're running with 2T command rate?
What's the white stuff on top of the tree in second render?
Yeah, it's preset like that with the XMP memory SPD profile in the BIOS, read off the ram. The ram is rated to run at 1600mhtz, 9-9-9-24 timings, 1.65 volts and 1.35 volt on the QPI/VTT
The white thing is the sun with partial development of aperture diffraction
You need MLT for proper aperture diffraction, but MLT is slower in this outdoor type scene than the new Path tracing code
I was experimenting with it yesterday
pretty cool, but I'm trying to emulate the same effect I see with my own eyes on bright light sources and specular reflections, where it's more random and a lot more off shoots. The smaller the aperture the more dramatic the effect.
caustics very slow to show in this scene, most probably due to the low sun position
Re: CoolColJ's test pics thread
Posted: Sun Jun 13, 2010 3:23 am
by dag
How did you get those really long flares? Tried 200.000 different settings.
Try obstacle maps for different effects.
Re: CoolColJ's test pics thread
Posted: Sun Jun 13, 2010 4:31 am
by CoolColJ
what you circled is the sun with a small bit of aperture diffraction, but it doesn't work well unless you use bidirectional MLT
As far as long flares go, after the experimentation I did yesterday, it's down to using a really smaller aperture size like f32- f64, and blade_curvature_radius of around 1, and a blade offset above 0.5
Both renders used similar values to below
f32 used here
<aperture_radius>0.000403</aperture_radius>
<aperture_shape>
<generated>
<num_blades>8</num_blades>
<blade_offset>0.600000</blade_offset>
<blade_curvature_radius>1.000000</blade_curvature_radius>
<start_angle>0.100000</start_angle>
</generated>
</aperture_shape>
The smaller the blade offset, the blurrier the image gets from testing. Values above 0.4 become clearer
I haven't tried any of the other stuff like obstacle maps etc yet.
above values with 16 blades, blade_offset of 0.01
0.1
0.2
Re: CoolColJ's test pics thread
Posted: Sun Jun 13, 2010 5:46 am
by dag
Blendigo only goes to 0.5 offset, but edited the igs with your settings. Still no really long flares here. Seems to be restricted to that square space.
What exporter do you use?
Can you post the igs without any objects, just the sun?
Re: CoolColJ's test pics thread
Posted: Sun Jun 13, 2010 10:54 am
by Ryouta Sumeragi
dag wrote:Blendigo only goes to 0.5 offset, but edited the igs with your settings. Still no really long flares here. Seems to be restricted to that square space.
What exporter do you use?
Can you post the igs without any objects, just the sun?
The memory doesn't suffice.
Or, if you are using 32 bit version, it is a specification......
Re: CoolColJ's test pics thread
Posted: Sun Jun 13, 2010 11:50 am
by dag
If you're referring to my RAM, I have 8 juicy gigs. I have blender 2.49b 64-bit, blendigo 2.4.3 and also indigo 2.4.4 64-bit.
Re: CoolColJ's test pics thread
Posted: Sun Jun 13, 2010 3:27 pm
by CoolColJ
Ok it's probably due to the rest of the camera values. I use C4D and Cindigo
I put 1 sphere in to give the camera something to focus on and make the tonemapper work properly
F64 was used here and gives longer flares. The smaller the aperture_radius the longer they become.
I think the rest of camera stuff like sensor width etc is always fixed.
Code: Select all
<renderer_settings>
<metropolis>true</metropolis>
<large_mutation_prob>0.330000</large_mutation_prob>
<max_change>0.010000</max_change>
<max_num_consec_rejections>1000</max_num_consec_rejections>
<max_depth>10000</max_depth>
<bih_tri_threshold>800000</bih_tri_threshold>
<halt_time>-1.000000</halt_time>
<halt_samples_per_pixel>-1.000000</halt_samples_per_pixel>
<bidirectional>true</bidirectional>
<hybrid>false</hybrid>
<logging>true</logging>
<display_period>60</display_period>
<image_save_period>300</image_save_period>
<save_untonemapped_exr>false</save_untonemapped_exr>
<save_tonemapped_exr>false</save_tonemapped_exr>
<save_igi>true</save_igi>
<auto_choose_num_threads>true</auto_choose_num_threads>
<cache_trees>false</cache_trees>
<polarisation>false</polarisation>
<super_sample_factor>2</super_sample_factor>
<splat_filter>
<mn_cubic>
<ring>0.333000</ring>
<blur>0.333000</blur>
</mn_cubic>
</splat_filter>
<downsize_filter>
<mn_cubic>
<ring>0.333000</ring>
<blur>0.333000</blur>
</mn_cubic>
</downsize_filter>
<watermark>true</watermark>
<info_overlay>true</info_overlay>
<width>800</width>
<height>600</height>
<post_process_diffraction>false</post_process_diffraction>
<aperture_diffraction>true</aperture_diffraction>
</renderer_settings>
<camera>
<aspect_ratio>1.333333</aspect_ratio>
<exposure_duration>0.050000</exposure_duration>
<pos>0.000000 0.000000 0.020000</pos>
<up>0.000000 0.000000 1.000000</up>
<forwards>-0.913545 0.406737 -0.000000</forwards>
<focus_distance>20.000000</focus_distance>
<sensor_width>0.027000</sensor_width>
<lens_sensor_dist>0.025933</lens_sensor_dist>
<aperture_radius>0.000203</aperture_radius>
<aperture_shape>
<generated>
<num_blades>16</num_blades>
<blade_offset>0.200000</blade_offset>
<blade_curvature_radius>2.000000</blade_curvature_radius>
<start_angle>0.100000</start_angle>
</generated>
</aperture_shape>
<white_balance>D65</white_balance>
<autofocus />
<lens_shift_up_distance>-0.000000</lens_shift_up_distance>
<lens_shift_right_distance>0.000000</lens_shift_right_distance>
</camera>
<skylight>
<sundir>-0.888126 0.393514 0.237443</sundir>
<turbidity>2.000000</turbidity>
<extra_atmospheric>false</extra_atmospheric>
<sun_layer>0</sun_layer>
<sky_layer>0</sky_layer>
</skylight>
<layer_name>
<layer_name>Sun/Sky</layer_name>
<layer_index>0</layer_index>
</layer_name>
<tonemapping>
<reinhard>
<pre_scale>10.000000</pre_scale>
<post_scale>1.000000</post_scale>
<burn>4.000000</burn>
</reinhard>
</tonemapping>
<material>
<name>defaultMat</name>
<diffuse>
<albedo>
<constant>
<rgb>
<rgb>0.800000 0.800000 0.800000</rgb>
<gamma>2.200000</gamma>
</rgb>
</constant>
</albedo>
</diffuse>
</material>
<sphere>
<!--Sphere-->
<center>-4.000000 1.662572 0.000000</center>
<radius>1.000000</radius>
<material_name>defaultMat</material_name>
</sphere>
</scene>
Re: CoolColJ's test pics thread
Posted: Sun Jun 13, 2010 3:33 pm
by CoolColJ
Real camera flares when using HDR tone mapping
Re: CoolColJ's test pics thread
Posted: Sun Jun 13, 2010 4:04 pm
by CoolColJ
Fstop 128 will give you really long flares for that sci fi look
<aperture_radius>0.000101</aperture_radius>
Re: CoolColJ's test pics thread
Posted: Sun Jun 13, 2010 6:34 pm
by Ryouta Sumeragi
In the PostProcess, a big flare becomes interrupted square.
It was heard that 64 bit version was normal.
(Because I do not have 64 bit environment, it is not possible to confirm it... )
My environment is WinXP32bit+4GB-RAM and Indigo2.4.4(Cindigo).
...Indigo is used only up to 2GB.
Re: CoolColJ's test pics thread
Posted: Mon Jun 14, 2010 3:48 pm
by CoolColJ
I don't use post process, it doesn't look as good.
The flares are not as large the last time I tried it
I used the obstacle map Dag posted earlier in this thread in this render. Looks pretty cool
More organic. Might try some finer dot obstacle map to simulate dust and imperfections on lens etc
now how to get these kinds of rings?
Re: CoolColJ's test pics thread
Posted: Tue Jun 15, 2010 12:23 am
by StompinTom
CoolColJ wrote:I don't use post process, it doesn't look as good.
The flares are not as large the last time I tried it
I used the obstacle map Dag posted earlier in this thread in this render. Looks pretty cool
More organic. Might try some finer dot obstacle map to simulate dust and imperfections on lens etc
im1276403182.jpg
now how to get these kinds of rings?
Those lens flare rings happen within the different glass elements of the lens. Until Indigo is actually simulating the lens elements, it won't happen. Somebody once made an actual camera with Indigo (modeled the lens pieces + plane to project the image onto) and got some reeeeally cool effects, though the render times were brutal of course.
Great tests!
Re: CoolColJ's test pics thread
Posted: Tue Jun 15, 2010 12:32 pm
by Godzilla
A few tests using Dag's diffraction map, hope you don't mind if I post them here aswell