JanHas Models & Skin Thread -Compatible with 4.36
-
All degrees of fukin’ awesomeness in this thread. Last set of screens gave me WOOD…!
PinocchioGorn
-
very heavy model, but nice work.
haha, the KF is the lightest poly model of the lot… All models so far are just below 100k… But I also made these models to last at least 10 years… by then it probably not a big deal anymore…
-
haha, the KF is the lightest poly model of the lot… All models so far are just below 100k… But I also made these models to last at least 10 years… by then it probably not a big deal anymore…
The Block 30/40/50… on your marks, get set, GO!
-
I see these pictures and I am so sorry I can’t install all your new works. With 4 block 52 in taxi I have 18 fps.
I need new rig.Congratulations for your work JanHas.
Even on my sig rig, I’ll often see 25-35 FPS in a mildy busy airbase. I think low FPS at taxi is typical.
If that CPU-Z summary refers to your rig I strongly recommend upgrading your CPU. With a killer GPU like a RX470 your CPU is bottle necking the entire rendering process. Even if you OC that Xeon to 4.0 or above it will always bottleneck the GPU. Also the relatively high multiplier for that manufacturing process era (I guess 32nm) doesn’t do you or the GPU any favor. Sandy- or Ivy-Bridge CPUs at 3.0GHZ would outperform your X5650. In fact even my Q6600 @ 3.35GHZ on air with an Asus STRIX does better. In this case the immense capability of your GPU actually worsens your problem hence the extremely low framerate you quoted. If that isn’t your rig ignore my comment.
No intention to hijack. JanHas your screenies and texture work are awe inspiring and I can’t quite believe it is Falcon whenever I see them so rock on!!!
Bottlenecks often caused by many factors. What we see very often in Falcon users is sudden FPS drops on the runway usually. Dx engine of the game must to be updated i know its hard but its old. Im getting also som fps drops and in Central europe test theater i dont want to mention what will happen if i install JanHas models. I dont want to think about it. All the objects and stuff and the models actually cause of the “old” DX engine are using too much memory i think. If we had a better engine that could utilize both nowadays cpu’s and gpu’s in 100% i think we would not have a problem even if the models where 300k tris or whatever.
@ JanHas Keep on rocking us with your superb work!!! -
Bottlenecks often caused by many factors. What we see very often in Falcon users is sudden FPS drops on the runway usually. Dx engine of the game must to be updated i know its hard but its old. Im getting also som fps drops and in Central europe test theater i dont want to mention what will happen if i install JanHas models. I dont want to think about it. All the objects and stuff and the models actually cause of the “old” DX engine are using too much memory i think. If we had a better engine that could utilize both nowadays cpu’s and gpu’s in 100% i think we would not have a problem even if the models where 300k tris or whatever.
@ JanHas Keep on rocking us with your superb work!!!All true I’m just saying that a relatively recent GPU can bottleneck older CPUs (even 6 core OC’d Xeons) quickly. That said I only claim validity on that statement based on my experiences in optimizing framerates of fairly recent heavy taxing games and engines where I’ve made this experience that a more recent CPU even at a lower clock will outperform older higher clocked ones in combination with a recent performance graphics card.
I absolutely can imagine that the old DX engine goes bezerk when you have 4 100k models in the same frame :).
-
JanHas. I’ve just download ALW v6. I’ve gone to install the DB110 Recon Pod and it has 4 LODs. I know how to change nLOD in LE, but your instructions say no parent data edit for this model. Could you tell me what the LOD distances should be please?
-
Usually better to email him amigo. It keeps this thread clear for his LVN 3 Viper models and also guarantees that he’ll see your question.
-
The KF-16V-52… V? Yes. V… is ready. Since the contract was signed last December to upgrade 134 KF-16’s to the V standard I took the opportunity to give it the V look. The center pedestal console, JHMCS and SniperXr are included. The AN/APX-125 upgrade is there as well. The model should be used to replace the KF-16C-52 model. The package can be downloaded via my website (janhas dot net). The next model to be build is the RNLAF F-16AM.
Just GORGEOUS Jan!!!
Keep on them…
Nikos. -
JanHas. I’ve just download ALW v6. I’ve gone to install the DB110 Recon Pod and it has 4 LODs. I know how to change nLOD in LE, but your instructions say no parent data edit for this model. Could you tell me what the LOD distances should be please?
As per instructions, use it to replace the MRP. The current MRP has only 1 lod and the Parent data jpeg says no edit needed, so replace LOD1 only. LOD2, 3 and 4 were made for the Ikaros theater and I left them in the package so other theater builders can use them if needed.
-
As per instructions, use it to replace the MRP. The current MRP has only 1 lod and the Parent data jpeg says no edit needed, so replace LOD1 only. LOD2, 3 and 4 were made for the Ikaros theater and I left them in the package so other theater builders can use them if needed.
Sweet. Yeah I was using it for the MRP. I’m all good then Thanks!
-
Appending new LOD number to LOD model allows you using LOD2,3 and 4.
-
Hi Jan, i tried only block 52 but in taxy …pull in the handle [emoji23][emoji23][emoji23] your work is good, but Falcon engine don’t love him. I download f16 k. Thanks for HD f16[emoji106]
-
I was wondering if switching hyper threading for off will improve the performance. I checked cpu (i7 3770k) performance with LE and saw it only use a single core (of the 8 cores). I have turned HT off and it improved LE (now uses 25%) with model saving times from 3,5 back to 2 minutes. I have read many threads about HT on other forums and it’s difficult to get a real answer. LE was written back in 2003 when HT did not exist, it’s not written to handle multiple cores (as is F4). Anyone have the same experience/idea about HT and F4?
-
I was wondering if switching hyper threading for off will improve the performance. I checked cpu (i7 3770k) performance with LE and saw it only use a single core (of the 8 cores). I have turned HT off and it improved LE (now uses 25%) with model saving times from 3,5 back to 2 minutes. I have read many threads about HT on other forums and it’s difficult to get a real answer. LE was written back in 2003 when HT did not exist, it’s not written to handle multiple cores (as is F4). Anyone have the same experience/idea about HT and F4?
I used to switch it off and feel better with F4. Don’t use many other sims or games, so can’t tell if it’s a general rule.
-
Hi Jan, i tried only block 52 but in taxy …pull in the handle [emoji23][emoji23][emoji23] your work is good, but Falcon engine don’t love him. I download f16 k. Thanks for HD f16[emoji106]
Best thing that can be done is to shrink the textures… 5-6 4096 textures for 1 F-16 model? This is way too much. Getting those down to 1024 or 2048 and putting all of them on a single 4096 sheet will for sure help. Sampling textures is still the most expensive operation done by the GPU and when you have many large textures, sure it’ll be more expensive.
There must be a correlation between texture size and the area that it’s going to cover… 4096 textures are maybe used by modern engines to tile very large geometry like e.g Terrain surfaces, but for a relatively small geometry like AC you usually get way too much oversampling in the pixel shader, so it’s just a waste of GPU time to load and sample such huge textures, and even more critical is the number of draw calls that need to be executed in order to draw 1 instance of the model (Because Falcon engine doesn’t support any kind of batching or instancing for 3D models).
-
Extensive tests in my old rig, using a full sq in the air, first time with the default models and second time with the models updated to lv3, proved an fps lose of around 3-4%, which IMHO some ppl might find funny vs what they will gain, in the eye-candy area of course.
As Ikaros theater is adding new stuff, I have scheduled some new tests exactly for this, and will be reported to theater notes.
-
…even more critical is the number of draw calls that need to be executed in order to draw 1 instance of the model (Because Falcon engine doesn’t support any kind of batching or instancing for 3D models).
You’re right but even with batching draw calls that would only benefit objects with the same mesh.
@ JanHas re hyperthreading:
The reason you don’t get a definitive answer is cuz there is none :). But you’re probably right that it benefits to leave it off.
Basically if it is enabled you need to make sure that your cores are ‘unparked’ first. Sth Win7 and Win 8.1 maybe even some 10 versions do when utilization falls below a certain threshold (~75%). It can be done by putting PC in high performance mode or with a third party tool. If all that is done and you still have better performance with HT off then the app in combination with the OS setting simply isn’t very good at utilizing all (or even more than 1) available cores. If that is the case even setting all cores to 100% at any given time will not improve anything at all.
When it comes to in-game performance everyone just has to check the CPU/GPU utilization in different situations (runway/taxi, midair, close to ground etc.) and on different cores. My guess is that F4 is mostly CPU bound hence the bottlenecking when you have a AAA graphics card calling on the CPU to deliver the frame while the CPU is still busy doing 15 other things that all depend on each other being finished BEFORE delivering the frame. If you know the utilization ratios then you can really start to optimize, before that it’s just guess work which can deliver improvements or actually make things worse.
Your textures which tax the system extraordinarily in combination with a CPU intensive application like Falcon makes for a great case study in performance optimization on newer multi-core multi-extension CPU/GPU systems with older programs like F4 no doubt!
-
Extensive tests in my old rig, using a full sq in the air, first time with the default models and second time with the models updated to lv3, proved an fps lose of around 3-4%, which IMHO some ppl might find funny vs what they will gain, in the eye-candy area of course.
As Ikaros theater is adding new stuff, I have scheduled some new tests exactly for this, and will be reported to theater notes.
define eye-candy area please
-
You’re right but even with batching draw calls that would only benefit objects with the same mesh.
Sure, that’s what I meant…
-
Jan your Block 42 made the news!