R Reynolds

Well I could say that it's only a WIP so the door has no hinge bones. But I also didn't I think of it.😄 I'm still on the learning curve of PowerDirector. Every time I convert a new AVI, I sometimes test a different compression setting just to see what happens. Thanks for the info.

Most images I post here are of a model in broad daylight. But since I'm building a railroad observation car I wanted to see if the internal lighting rig would be believable from the outside. The only way to check it would be to put the model in subdued light, so here it is just after sundown. obsv_car_walk_around_01.mp4

Here's my observation car entering the concourse at too high a speed with a too abrupt stop. I have a low opinion of moving your virtual camera through a glass window but the reason for this test shot was to check that the car's interior ambient light rig balanced well with that of the concourse. obsv_car_entrance_01.mp4

True, but intensity tests I made showed that a global ambience bright enough to get a good exposure inside the car totally over exposes the outside surroundings as seen through the windows. What I'm going for is a universal lighting setup so the camera can show a character sitting inside the car looking out as the train pulls into the station. Then, in a continuous move, the camera can follow that character as he gets off and walks along the platform with all intensities being believable. You've mentioned this before concerning images of my large train station model. I seem to remember Yves writing that radiosity only works in small enclosed spaces. Once a character walks into the giant rooms of the station or for that matter "outside" under a 1 mile (1.6 km) diameter sky dome, I assumed all bets were off.

My best solution so far is to set the darkness of both the white sun and concurrent negative blue sun darkness to 100%. Outside I can achieve believable ambient fill with global ambient intensity at 140% from the surrounding sky dome. However inside an enclosure, not enough global ambient rays enter to produce adequate exposures and I have to crank occlusion sampling to 100% to minimize the noise it generates on surfaces. So to get a reasonable exposure inside something like the WIP observation car (shown attached) I put a properly tweaked klieg outside each window pointed inward. The passenger car is made of two models, exterior and interior. These ambient skylight kliegs get put into a light list so only the interior model is lit by them. I'm not sure I've got the right light levels but I'm getting close. passenger_car_pass_thru_00.mp4

I've been "bit" more than once by the way the Darkness attribute works when a light cast shadows. I know why the variable is there; it's a way to fake ambient light filling in the shade. But I have a problem with the way it affects specular highlights. It still lets them through, just attenuated. The specular highlight is faking the reflection of the light source in the sphere. Even if the shadow darkness is 10%, that "reflection" still shouldn't be visible. I have no idea whether it's possible for Stefen to fix this in the code but it would be a nice improvement.

From your 2016 reference; Since all my hardware runs Windows 10, I've hit a wall.

I have a vague memory that the programmer's name came up when you opened the plugin. Can you still install v13 (2005 "Sorserez" disk) and run the hxt? My hat's off to the both of you. The last time I coded anything in C was in the last century and my brain would always hurt after someone tried to make me understand the concept of a pointer to an array of pointers.

While rummaging through an old miscellaneous folder I found a plug-in from 2005 that, if I remember correctly, simultaneously changes the crossing bias at all cp's around a circular spline. It could be a nice time saver. I tried dropping it into my HXT folder but it didn't show up as a plug-in or wizard. Is there some way to "re-compile" it for v19p? Bottleneck.hxt

Yeah, it would never work for the car that needs to be ready for its close-up but for random vehicles parked in the background, more than adequate.

I've discovered a progressive render issue in v19.0p. While it's not a show stopper it's been annoying me for a long time and I've finally found the root cause. This image is an overview of the test chor. There's a camera with a 1000mm lens looking at five instances of a grey reference sphere, each at a different distance from the camera. The sphere instances are scaled in the chor as they get closer to the camera so that from the camera view they don't change too much in image size. They are positioned so the camera can see all five spheres side by side. Between the camera and the closest sphere is a slightly tinted, four patch plane that is a "glass sheet" with a transparency of 80%. It is positioned so that the camera only sees the lower half of the spheres through it. A simple sun light shines down from above and the chor has global ambience turned on with a white colour; intensity = 80% and occlusion = 100%. This image is the results of 5 pass prog renders and 9 pass full screen renders. The first two sphere rows are with no glass in the way while the last two rows have the glass turned on. For prog. renders through a transparent surface something strange happens as the surface gets farther from the camera. If you turn off ambience and leave in the glass slide this issue also goes away. Although it always happens in progressive renders, I have seen it occasionally in 9 pass full screen renders of an 80Mb project chor. So if Stefan does try to fix this issue he may want to look at both types of screen rendering. amb_occl_thru_glass_distant_embed.prj

That's really impressive; especially the deep cutout for the license plate which I never would have expected to be that believable. Well done!

Nicely done. Wireframe of the vase, please? I'm curious how you ended the vee grooves at the neck.

Now that I've finally optimized the settings for klieg lights as window skylights, I'm getting closer to having an interior light level that's believable. I'm still tweaking intensities to get good exposure levels from full sun to inside the archway but this is an improvement on the lighting in the Oct. 1 image of this arch since we can now see the spill light from the concourse beyond the doors.

I've practically spent the last quarter of 2021 on the arched concourse that can be seen through the archway in the last image. As I said at the beginning of this thread, the real N.Y. Penn Station concourse was an elaborate waiting room for travelers using the stairways to the trains in basement tunnels. Great architecture but I wanted something more along the line of what the Pennsylvania Railroad built in Pittsburgh, a large train shed. Basically a structure built to accommodate multiple, entire passenger trains. But I couldn't imagine the amount of my time and computer RAM it would take to model that intricate array of beams and girders so I decided to go with a complete fantasy design of riveted box arches. This design means far fewer patches and the rivets and stamped patterns can be faked using normal maps. And here's a short camera move shooting inside, near a painfully under-modeled lounge/observation car which is here mostly to give a sense of the scale of the building. walk_thru_00.mp4 I'm beginning to get the feeling that I may never get to the end of the line on this project but I'm having a blast during the trip.