c172p icon indicating copy to clipboard operation
c172p copied to clipboard

Jagged lines in reflection in glass

Open legoboyvdlp opened this issue 7 years ago • 44 comments
trafficstars

image

image

On branch bug-1015

legoboyvdlp avatar Mar 20 '18 18:03 legoboyvdlp

What's wrong with that? Just some reflection distorted by the glass, isn't it?

gilbertohasnofb avatar Mar 20 '18 18:03 gilbertohasnofb

I'm a little concerned about the jagged lines. What exactly is it reflecting there? But perhaps the new reflection cubemap might reduce the severity of those lines?

legoboyvdlp avatar Mar 20 '18 18:03 legoboyvdlp

What exactly is it reflecting there?

Some random image that is not a true reflection of the world outside.

But perhaps the new reflection cubemap might reduce the severity of those lines?

I don't think that's worth it to be honest. @wlbragg?

gilbertohasnofb avatar Mar 20 '18 19:03 gilbertohasnofb

I don't remember exactly where I pointed it out, but I am not totally happy with the reflection cubes we are using, Problem is I don't know that we can "easily" improve upon them. They are Blender generated and are reflecting what you see in Blenders Model view, not real life (or in this case in sim). Because of the way models were historically put together they also don't contain any of the instruments. So the reflections aren't really even close but a distant facsimile. My suggestion is someone try to create new maps from in the sim VS in Blender. The issue there is the scale is not square. But if you could start the sim in the most square resolution you can, and basically crop the screen shot square, maybe we could achieve satisfactory results. I think they would look better with missing data (because of the dimensions and having to square crop) than full data cubes but using a facsimile of the rendered visual. I could be wrong, but I think it would look more realistic. As far as this glass distortion were seeing now, I don't know that that will change much because it is being viewed at such an extreme angle, I think.

wlbragg avatar Mar 20 '18 19:03 wlbragg

They are Blender generated and are reflecting what you see in Blenders Model view

Are you sure about that? I took a look at them and they seemed to be using FG textures and everything. There are a bunch of cube maps in that FGDATA folder, but all of them look equally bad. :cry:

gilbertohasnofb avatar Mar 20 '18 23:03 gilbertohasnofb

No. you right @gilbertohasnofb I was looking at the backup colored maps, specifically the red one... red_nx However the ones were using do appear to be rendered from in the sim. reflection_nx

wlbragg avatar Mar 21 '18 01:03 wlbragg

Although they do appear to be too dark in my opinion. I think we may be losing some detail. Also why the gray windows VS white? This was the original grayscale version that wasn't proportionally correct. grayscale_nx

wlbragg avatar Mar 21 '18 01:03 wlbragg

The first image are the old wrong proportioned gray scale versions.. fgfs-screen-006 fgfs-screen-007 These are what we are using now... fgfs-screen-004 fgfs-screen-005

wlbragg avatar Mar 21 '18 01:03 wlbragg

@wlbragg I think the reason we are using grey scale cockpit maps is because we have many different liveries for the cockpit but a single map, and back in the day we decided that a BW map would fit better with all of them. I still think that's an all right decision as the reflection is rarely that noticeable (in the sense of being able to really see cockpit parts and other things in the reflected image)

gilbertohasnofb avatar Mar 24 '18 16:03 gilbertohasnofb

@gilbertohasnofb

I think the reason we are using grey scale cockpit maps is

I'm not questioning that, I agree.

Now my only "potential" issue is the gray scale maps we are using now. They have a gray (or darker) window area that I think should really be white? The area that is window is actually gray in color and I don't think that is right.

Also, as a whole they appear to be a little darker than the original "gray scale" ones we used to use. Those we replaced because they were distorted proportionally.

wlbragg avatar Mar 24 '18 16:03 wlbragg

I see, I will take a look at that.

gilbertohasnofb avatar Mar 24 '18 16:03 gilbertohasnofb

@wlbragg done, reflection maps should have a white background, could you please confirm if that improves the issue?

gilbertohasnofb avatar Mar 24 '18 17:03 gilbertohasnofb

@wlbragg

image

Is that line meant to be half green and half white on the back window?

legoboyvdlp avatar Mar 26 '18 14:03 legoboyvdlp

Is that line meant to be half green and half white on the back window?

No, but that's a livery mistake. I will fix that.

gilbertohasnofb avatar Mar 26 '18 21:03 gilbertohasnofb

@legoboyvdlp I fixed this in the opened damage model PR branch refactor-damage-model

gilbertohasnofb avatar Mar 26 '18 21:03 gilbertohasnofb

Further problem: image

legoboyvdlp avatar Mar 27 '18 17:03 legoboyvdlp

Looks to me that the ice formation is mapped just frontally and so it looks terrible on the sides.

@legoboyvdlp turn off that HUD and fly this beauty using its own instruments!! :rofl:

gilbertohasnofb avatar Mar 27 '18 17:03 gilbertohasnofb

We should totally ban the HUD :stuck_out_tongue_winking_eye:

onox avatar Mar 27 '18 18:03 onox

Can we verify the ice is not working on any sides? Also be aware of the fact that the front windshield is one piece and so if the front windshield has ice then the side is also included in the effect. This may have to do with the effects switch used to distinguished between the two different angles of glass. The more horizontal VS vertical glass.

We need to revisit this whole fog, rain, reflection effect and make sure we didn't inadvertently screw it up with the changing of the windshield.

I would like to see some good imaged of ice on all the windows.

wlbragg avatar Mar 27 '18 19:03 wlbragg

We should totally ban the HUD stuck_out_tongue_winking_eye

Someone please make a PR for the main FG project removing that thing! :rofl:

gilbertohasnofb avatar Mar 27 '18 23:03 gilbertohasnofb

@wlbragg Sorry for my ignorance, but how can I force the ice to appear on the glass?

But it does look the modifications to the glass made the effects a bit strange, the rain also looks distorted on the sides:

fg

gilbertohasnofb avatar Mar 27 '18 23:03 gilbertohasnofb

But it does look the modifications to the glass made the effects a bit strange, the rain also looks distorted on the sides:

Yes, that is a fact and there is no good solution to correct that. The reason is we have two modes to do the rain in the effect

<surface-mapping-scheme type="int">1</surface-mapping-scheme>

and

<surface-mapping-scheme type="int">2</surface-mapping-scheme>

Mode 1 is for more horizontal glass faces and mode 2 is for the vertical glass faces. Because our front window is one object now we can only assign mode 1. We used to assign mode 2 to the side objects. We still assign mode 2 to the door and aback seat side windows.

By the way, I just pushed the Blend and .ac turning on some objects I had off while doing the livery work. I forgot to turn them back on.

As far as the ice is concerned, I'll have to experiment before I can suggest a method to easily see it.

wlbragg avatar Mar 28 '18 00:03 wlbragg

@gilbertohasnofb I just started from KEQA, I had the aircraft started up and sitting on the runway at startup. It was sitting there for a little while and the I ctrl+U up to maybe 10-12 thousand feet. I opened the heater slider all the way for a minute or two, then opened the air slider all the way and it started to show the ice pretty quickly. I then shut both sliders and paused the simulation. fgfs-screen-008

As you can see it is all working but indeed the windshield side glass is not looking good because of it being only one object and having to be assigned mode 1. The only solution is to separate the sides to their own objects like it was before so we can assign mode 2 to them (and put up with the hard edge) .

wlbragg avatar Mar 28 '18 00:03 wlbragg

@wlbragg I see, thanks for the detailed post above.

The only solution is to separate the sides to their own objects so we can assign mode 2 to them.

What would be the side effect of doing that? Would that reinstate the issue of the glass looking a bit weird in the cockpit, which is what we had before?

gilbertohasnofb avatar Mar 28 '18 00:03 gilbertohasnofb

Meaning, would it look again like this? https://github.com/c172p-team/c172p-detailed/issues/1010

gilbertohasnofb avatar Mar 28 '18 00:03 gilbertohasnofb

Yes!

wlbragg avatar Mar 28 '18 00:03 wlbragg

Hmm, so we have to choose between pretty glass vs pretty effects. What would be your preference? I am really unsure myself. The effects do not look good, but those are visible on 1% of flights while the front glass is always there. On the other, the divided glass does not look that bad and one can argue that that's totally acceptable in order to have the effects working well. What would you pick?

gilbertohasnofb avatar Mar 28 '18 00:03 gilbertohasnofb

Here it is as before fgfs-screen-009 fgfs-screen-011

wlbragg avatar Mar 28 '18 00:03 wlbragg

The effect certainly looks better, but the glass does look a bit worse. We need a choice: what would you pick?

gilbertohasnofb avatar Mar 28 '18 00:03 gilbertohasnofb

For completeness, here it is now. fgfs-screen-014 fgfs-screen-013

I guess with the hard edge? I would like @onox and @legoboyvdlp to weight in!

wlbragg avatar Mar 28 '18 00:03 wlbragg