Would like to read thoughts on issue w/Thrustmaster TX Leather edition
Hello simracers! Reposting here from simracing, just trying to reach a few more people.
Few days ago saw a really good deal for a used set of a Thrustmaster TX Leather, a TH8A and T-LCM pedals. I've had had mostly good experiences with buying used hardware and I have a brother who'd make good use of my old trusty G29.
So I pulled the plug and bought it. Once I got it setup on the Playseat I had a go with it. Everything seemed good, wheel rotation feels smooth, pedals feel miles ahead of the G29s, and the shifter is sublime! No input spiking (damn Logi potentiometers)
I knew these Thrustmaster wheels have had spotty QC but from what I've read its stuff that pops up early and in warranty. Seller said he's done some AC drifting but not much more. Gave me an impression of a guy who got it to have fun driving during the COVID downtime.
Have already done a few 30-40 minute sessions in EA WRC, FM8 and ACC, no overheating and fading FFB.
But there is one issue - for whatever reason the wheel doesn't go to full 450 degrees rotation to the right. The autocalibration does its thing and the wheel is centered perfectly but as I turn the wheel to the right it starts do to slightly more degrees of rotation in Windows and ends up hitting the endstop at about I would guess 420-30 degrees. I did do the driver uninstall and reinstalled, tried multiple different rear case USB ports as well as getting the latest firmware update but didn't help. I do notice that when the autocalibration does its thing, it seems to slightly slow down at the full right end stop. If my description of the issue isn't clear enough check my last couple of comments on Reddit
Has anyone here had this issue? If yes, has it gotten worse? Is it a sign of the wheelbase failing? Is it some bizarre software issue?
Has anyone had experience disassembling these wheelbases? If so would you have an an idea of what this could be? I have done some work on my Logitech pedals in the past but nothing major, just some potentiometer cleaning and minor spring adjusting.
Do I just bite the bullet and live with it? I'm far from a hardcore simracer, I don't run leagues or leaderboards. To be honest, I don't think I would've even noticed this in game just by regular driving. The pedals and the shifter I absolutely love. The FFB itself too, feels like a massive upgrade too over my previous Logi wheels.

Hello simracers!
Few days ago saw a really good deal for a used set of a Thrustmaster TX Leather, a TH8A and T-LCM pedals. I've had had mostly good experiences with buying used hardware and I have a brother who'd make good use of my old trusty G29.
So I pulled the plug and bought it. Once I got it setup on the Playseat I had a go with it. Everything seemed good, wheel rotation feels smooth, pedals feel miles ahead of the G29s, and the shifter is sublime! No input spiking (damn Logi potentiometers)
I knew these Thrustmaster wheels have had spotty QC but from what I've read its stuff that pops up early and in warranty. Seller said he's done some AC drifting but not much more. Gave me an impression of a guy who got it to have fun driving during the COVID downtime.
Have already done a few 30-40 minute sessions in EA WRC, FM8 and ACC, no overheating and fading FFB.
But there is one issue - for whatever reason the wheel doesn't go to full 450 degrees rotation to the right. The autocalibration does its thing and the wheel is centered perfectly but as I turn the wheel to the right it starts do to slightly more degrees of rotation in Windows and ends up hitting the endstop at about I would guess 420-30 degrees. I did do the driver uninstall and reinstalled, tried multiple different rear case USB ports as well as getting the latest firmware update but didn't help. I do notice that when the autocalibration does its thing, it seems to slightly slow down at the full right end stop. If my description of the issue isn't clear enough check my last couple of comments on Reddit
Has anyone here had this issue? If yes, has it gotten worse? Is it a sign of the wheelbase failing? Is it some bizarre software issue?
Has anyone had experience disassembling these wheelbases? If so would you have an an idea of what this could be? I have done some work on my Logitech pedals in the past but nothing major, just some potentiometer cleaning and minor spring adjusting.
Do I just bite the bullet and live with it? I'm far from a hardcore simracer, I don't run leagues or leaderboards. To be honest, I don't think I would've even noticed this in game just by regular driving. The pedals and the shifter I absolutely love. The FFB itself too, feels like a massive upgrade too over my previous Logi wheels.

Hello! I've been playing with my TS-XW + T-LCM on my PC for quite some time now and had plenty of fun. A few days ago I've finally decided to buy myself a TSS Handbrake and a TH8A shifter. I've installed all the drivers, plugged both devices and... Nothing, my game (Dirt Rally 1) just wouldn't read them. I've opened up the thrustmaster's control panel only to find that my wheel is the only device that appears there.
The handbrake appeared when I connected it to my PC via the DIN/USB cable. I had previously plugged it into the servo base so that was probably the problem. I only have ONE DIN/USB cable that was included with the TH8A! From what I understand I need to buy another cable in order to get the handbrake to work? Please correct me if I'm wrong. The shifter simply won't appear in the control panel no matter what I do. I have it connected to my PC via the only DIN/USB cable I've received, tried multiple different USB ports with no avail. When I boot up the calibration software it just looks as if it's bugged or incomplete. What should I do now? I've already contacted the Thrustmaster support but I would also like to look for help in here. The shifter is listed as an installed "app" on my PC. When I plug it there's no noise indicating that it has been plugged.
Please excuse me if something I wrote in here is hard to understand, I'm currently VERY sick and also English is not my first language.
EDIT: Turns out I didn't get a DIN/USB Cable in my Handbrake box, my vendor says it's pretty common and they sent me one for free!
Control panel
The blank calibration software

We have some old lpe1250 FC adapters, I can't create a proper image in vLCM with last working driver 12.8.61 and the regular Fujitus Vendor AddOn. Now I want to create a customized offline bundle wit the Fujitsu zip + lpfc zip that I can then use. For that I cloned the Fujitsu offline bundle.
Is there any way to add lpfc 12.8.614.16 to a new image? The Add-EsxSoftwarePackage seems not to have the option to choose from a version. I tried Remove-EsxSoftwarePackage for the lpfc driver but after adding "lpfc" again there

# Get-EsxSoftwareDepot Depot Url --------- zip:D:\temp\Broadcom-ELX-lpfc_12.8.614.16-1OEM.670.0.0.8169922-offline_bundle-19480864.zip?index.xml file://custom/depot/index.xml zip:D:\temp\VMware-ESXi-7.0.3.update03-21686933-Fujitsu-v531-1-offline_bundle.zip?index.xml # Get-EsxImageProfile Name Vendor Last Modified Acceptance Level ---- ------ ------------- ---------------- My_old-lpfc_FJT-Addon-fo... Fujitsu 16/05/2024 1... PartnerSupported FJT-Addon-for-FujitsuCustom... Fujitsu 22/05/2023 0... PartnerSupported # Get-EsxSoftwarePackage lpfc Name Version Vendor Creation Date ---- ------- ------ ------------- lpfc 14.0.326.12-1OEM.700.1.0.15... EMU 05/10/2021 18... lpfc 14.0.169.26-5vmw.703.0.50.2... VMW 30/06/2022 14... lpfc 12.8.614.16-1OEM.670.0.0.81... EMU 25/02/2022 11... 

Hi, I have a Asus ROG Strix G17 (2022) G713 with this panel: https://www.panelook.com/NE173FHM-NZ6_BOE_17.3_LCM_overview_52637.html
I found this panel that looks good it has mini-led + hdr: https://www.panelook.com/modeldetail.php?id=54254
I want to upgrade my screen with that panel, it the same connector and same voltage but different input current is 660mA vs 1.07A, also there is PWM dimming with led driver but my current screen don't have PWM dimming. Can I upgrade to the new panel ?

Hi, I have a Asus ROG Strix G17 (2022) G713 with this panel: https://www.panelook.com/NE173FHM-NZ6_BOE_17.3_LCM_overview_52637.html
I found this panel that looks good it has mini-led + hdr: https://www.panelook.com/modeldetail.php?id=54254
I want to upgrade my screen with that panel, it the same connector and same voltage but different input current 660mA vs 1.07A, also there is PWM dimming with led driver but my current screen don't have PWM dimming. Can I upgrade to the new panel ?

Hi, I have a Asus ROG Strix G17 (2022) G713 with this panel: https://www.panelook.com/NE173FHM-NZ6_BOE_17.3_LCM_overview_52637.html
I found this pannel that look good with mini-led + hdr: https://www.panelook.com/modeldetail.php?id=54254
I want to upgrade it and found this one with the same connector and same voltage but different input current 660mA vs 1.07A also there is PWM dimming with led driver but my current screen don't have PWM dimming. Can I upgrade to the new pannel ?

From Bike Durham yesterday:
I have tough news to share. [Monday] night and early [yesterday] morning, two people were hit by drivers while walking. One man is in critical condition after a driver hit him and left the scene last night at North Miami Boulevard near Raynor Street. Early [yesterday] morning, another man was killed after he was hit by a driver who fled on North Roxboro Street, just north of Club Boulevard. Both of these tragedies occurred on state-maintained streets where the most severe traffic violence occurs due to higher driving speeds.
This stretch of North Roxboro Street is included in the Durham Bicycle and Pedestrian Advisory Commission's recent letter to City Council, County Commission, and NCDOT requesting a conversion of North Roxboro Street from 4 lanes to 3. According to crash data, this is the most dangerous section of North Roxboro, and one of the most dangerous streets in the city. We support their recommendations as part of a goal to make the entire length of Roxboro Street safe for all users. The section of Roxboro and Mangum where we've been pushing for a conversion to two-way traffic is just the first step in addressing the entire corridor.
Speaking of the section of Roxboro and Mangum, on April 26, the Indy published an article that first ran in the 9th Street Journal entitled "Reckless Roxboro May Get a Revamp if Locals Can Persuade NCDOT". I encourage you to read the article by Esmé Fox and pay attention to the troubling quotes by NCDOT district engineer John Sandor. You can read my full response to this article, as well as the letter to the editor that I submitted at today's blog post.

What You Can Do

1. Read today's blog post. Get familiar with this issue.
2. Send your own letter to the editor at IndyWeek ([backtalk@indyweek.com](mailto:backtalk@indyweek.com?subject=Response+to)) in response to "Reckless Roxboro May Get a Revamp if Locals Can Persuade NCDOT". The limit is 300 words.
3. Sign up to receive email updates on our Safe Streets campaign.
4. Become a sustainer to power this campaign for safe streets by making a monthly or annual contribution.

Be well,
John

* This is for Windows only , because zluda is windows only.
[[ https://github.com/patientx/ComfyUI-Zluda ]]
-You just need to install AMD Hip Software then you are good to go.
-Just clone - run install.bat and double / triple / quadruple your generation speed. Also far less memory problems and faster load times.
For comparison : I have an RX 6600 8 GB and with the card at its normal daily use (no overlocking or undervolting (100w according to driver))

::: sd1.5 :::	::: sdxl :::	sdxl with deepcache
1.7sec/it	10sec/it	2.38sec/it
2.95it/sec	2.32sec/it	1.32it/sec

* Yes, deepcache itself makes significant difference BUT it degrades the quality so you have to do more steps , and it is not very compatible with other model altering stuff.
* You should use it and double/triple your step count , still faster than not using it and check the final quality.
* Of course you can go faster if you use lcm / lightning / hyper loras in your workflow. These are just a comparison for standard use.
* The only downside of using ZLUDA so far I've found is, since CDUNN isn't supported facerestoration using codeformer doesn't work because it uses onnx and that uses cdunn, a workaround for that I use is instantid or do facerestoration with maybe facefusion or similar outside app if you are that desperate. Also some other nodes use onnx and or other packages that rely on cudnn they are a no-go here too. I didn't find any that I use constantly though.
** Have to add: This works with amd gpus from 6800 and up natively and there are optimized libraries that you can use for all 6600 / 6700 cards , and these (gfx803 ! gfx900 gfx906 ! gfx1010 gfx1011 gfx1012 gfx1030 gfx1031 gfx1032 gfx1100 gfx1101 gfx1102 ) at https://github.com/brknsoul/ROCmLibs . Don't ask / bother if they work please, just test it yourself, it seems a few of them are already tested.

This post contains my working printer.cfg for the Geeetech A10 Pro 3D Printer. I am posting this due to the lack of other resources on converting the GT2560 v4.1B board inside of the printer to Klipper. If you have the same Printer or board you should be able to use this printer.cfg directly or as a base for your own config.
I recently got this old used Geeetech A10 Pro Ender-3 clone. this printer ships with a special low cost variant of Geeetech's GT2560 board. unlike the non Pro variant this one has many missing connectors and non replaceable hard soldered A4988 drivers (also only 4 of the 6 positions are actually used).
the Klipper Github has an example config for a generic GT2560 board but that config has most of the pins assigned wrong for this model.
i found a schematic of the v4.1B board on the Geeetech website (downloads section -> A10 -> GT2560V4.1BSCHA10.pdf) and built a printer.cfg based on it. My printer has a BL Touch installed so it is configured to use that instead of the regular z end stop. i also tried to get the display working but even after assigning the pins it did not work correctly. everything else works fine though so i want to share my cfg here for people with the same board who want to run Klipper:
the config includes notes as to what the pin is named in the schematic so it is easier to understand the assignments. of course you have to change the mcu serial and virtual sd card path to what you have set on your own klipper install.

[include fluidd.cfg] [include mainsail.cfg] [mcu] serial: [YOUR MCU PATH] [virtual_sdcard] path: [YOUR VIRTUAL SD PATH] on_error_gcode: CANCEL_PRINT # This file contains common pin mappings for the Geeetech GT2560 v4.1B # board. GT2560 board uses a firmware compiled for the AVR # atmega2560. # Schematic Download: https://www.geeetech.com/download.html?version_id=446 # See docs/Config_Reference.md for a description of parameters. [stepper_x] step_pin: PC0 #in schematic: X_STEP dir_pin: !PG2 #in schematic: X_DIR enable_pin: !PC2 #in schematic: X_EN microsteps: 16 rotation_distance: 40 endstop_pin: !PA2 #in schematic: X_MIN position_endstop: 0 position_max: 235 homing_speed: 30 [stepper_y] step_pin: PC6 #in schematic: Y_STEP dir_pin: !PC4 #in schematic: Y_DIR enable_pin: !PA7 #in schematic: Y_EN microsteps: 16 rotation_distance: 40 endstop_pin: !PA6 #in schematic: Y_MIN position_endstop: 0 position_max: 225 homing_speed: 30 [stepper_z] step_pin: PA3 #in schematic: Z_STEP dir_pin: PA1 #in schematic: Z_DIR enable_pin: !PA5 #in schematic: Z_EN microsteps: 16 rotation_distance: 8 #endstop_pin: ^PC7 #in schematic: Z0_MIN endstop_pin: probe:z_virtual_endstop #position_endstop: 0 position_max: 235 position_min: -5 [bltouch] sensor_pin: PC7 #in schematic: Z_MIN1 / Z0_MIN control_pin: PB5 #in schematic: PB5 / PB5A x_offset: -45.4 y_offset: -7 #z_offset: 0 speed: 20 samples: 1 sample_retract_dist: 8 [bed_mesh] speed: 150 mesh_min: 10,0 mesh_max: 185,200 probe_count: 8,8 fade_start: 1 fade_end: 10 fade_target: 0 algorithm: bicubic [safe_z_home] home_xy_position: 155.4,117 speed: 50 z_hop: 10 z_hop_speed: 5 [extruder] step_pin: PL3 #in schematic: E0_STEP dir_pin: !PL5 #in schematic: E0_DIR enable_pin: !PB6 #in schematic: E0_EN microsteps: 16 rotation_distance: 33.683 nozzle_diameter: 0.4 filament_diameter: 1.750 heater_pin: PB4 #in schematic: PB4_HE2 #sensor_type: ATC Semitec 104GT-2 sensor_type: EPCOS 100K B57560G104F sensor_pin: PK3 #in schematic: T0 min_temp: 0 max_temp: 250 control: pid pid_kp: 29.800 pid_ki: 1.774 pid_kd: 125.159 [heater_bed] heater_pin: PG5 #in schematic: PWM3_HOTBED sensor_type: ATC Semitec 104GT-2 sensor_pin: PK2 #in schematic: T_BED min_temp: 0 max_temp: 120 control: pid pid_kp: 63.041 pid_ki: 2.898 pid_kd: 342.787 [fan] pin: PH6 #in schematic: PWM8_FAN0 [printer] kinematics: cartesian max_velocity: 200 max_accel: 1500 max_z_velocity: 20 max_z_accel: 500 [display] lcd_type: hd44780 rs_pin: PD1 #in schematic: LCM_RS e_pin: PH0 #in schematic: LCM_EN d4_pin: PH1 #in schematic: LCM_D4 d5_pin: PD0 #in schematic: LCM_D5 d6_pin: PE3 #in schematic: LCM_D6 d7_pin: PC1 #in schematic: LCM_D7 encoder_pins: PL7, PG1 click_pin: PD2 #in schematic: EC_PRESS [gcode_macro G29] gcode: G28 BED_MESH_CALIBRATE G1 X0 Y0 Z10 F4200 [gcode_macro Bed_Mesh_Calibrate] gcode: G28 BED_MESH_CALIBRATE G1 X0 Y0 Z10 F4200 [gcode_macro Z_Offset_Calibrate] gcode: G28 PROBE_CALIBRATE [gcode_macro Save] gcode: SAVE_CONFIG #*# <---------------------- SAVE_CONFIG -------------------- #*# DO NOT EDIT THIS BLOCK OR BELOW. The contents are auto-generated. #*# #*# [bed_mesh default] #*# version = 1 #*# points = #*# -0.212500, -0.325000, -0.412500, -0.432500, -0.505000, -0.532500, -0.607500, -0.707500 #*# -0.055000, -0.140000, -0.210000, -0.222500, -0.292500, -0.332500, -0.410000, -0.515000 #*# -0.025000, -0.120000, -0.185000, -0.237500, -0.320000, -0.392500, -0.492500, -0.635000 #*# 0.287500, 0.175000, 0.067500, 0.012500, -0.115000, -0.245000, -0.392500, -0.602500 #*# 0.450000, 0.317500, 0.185000, 0.112500, -0.007500, -0.130000, -0.280000, -0.452500 #*# 0.720000, 0.572500, 0.435000, 0.335000, 0.195000, 0.080000, -0.100000, -0.312500 #*# 0.980000, 0.812500, 0.655000, 0.515000, 0.355000, 0.192500, 0.042500, -0.225000 #*# 1.085000, 0.915000, 0.767500, 0.635000, 0.465000, 0.335000, 0.167500, -0.065000 #*# x_count = 8 #*# y_count = 8 #*# mesh_x_pps = 2 #*# mesh_y_pps = 2 #*# algo = bicubic #*# tension = 0.2 #*# min_x = 10.0 #*# max_x = 185.0 #*# min_y = 0.0 #*# max_y = 199.99 #*# #*# [bltouch] #*# z_offset = 2.505 

Hi All,
I am looking for a driver board that will work with this lcd panel: vvx10f004b00 (https://www.panelook.com/VVX10F004B00_Panasonic_10.1_LCM_overview_19637.html), salvaged from Asus Transformer Pad Infinity 700 - tf700t (https://www.gsmarena.com/asus_transformer_pad_infinity_700-4606.php).
The panel is really good - Full HD with good density and can be quite usefull to use with Raspberry PI, so if someone knows how can I get to hook it up to a commonly used interface such as VGA, HDMA or other, please do let me know.
Thanks!

We have a few old server with lpe1250 HBA. They don't work with any current lpfc driver, last supported version is lpfc_14.0.169.26-5. I just switched from baseline to image and this installed a driver from Fujitsu vendor addon that was too new and now no HBAs/LUNs are visible. My mistake.
I can manually install the old VMW_bootbank_lpfc_14.0.169.26-5vmw.703.0.50.20036589.vib (which was hard to find as there is no link anywhere as this an inbox driver that came with 7.0 U3).
But how can I add this version to the vLCM image? I can't add the file VMW_bootbank_lpfc_14.0.169.26-5vmw.703.0.50.20036589.vib to the repository. And this lpfc version is not available in the dropdown for lpfc package during image creation/modification. Only older 12.8 versions or newer 14.x versions. How can I create a valid image now?
https://i.imgur.com/7KRD8Ec.png

I own a Lenovo V14 G3 IAP Laptop - Type 82TS with a B140HTN02.1 Screen. I found this LG Sreen (LP140WF5-SPK1) that has a Touchscreen. But I don't know what driver I will need. On top of that I'm not sure if my Laptop supports Touch.

I'm wanting to use LCM to update a cluster,but I'm getting these two errors:

Identified standalone vib(s) ntg3 4.1.9.0-5vmw.703.0.90.21686933 belonging to Broadcom NetXtreme I ESX VMKAPI ethernet driver 4.1.9.0-5vmw component.Review the compliance details and ensure no warnings about these vibs are present. If present, add given component to the image to include these vibs.
Identified standalone vib vmware-fdm 6.7.0-15132721 on the following host(s): hpk-vmw-esxeng1.digi.comReview the compliance details and ensure no warning about this vib is present. If present, import the component containing the given vib and add it to the image.

I'm curious what you all do with these? Do they concern you at all? If so, what do you do about it before allowing LCM to do the updates?
I appreciate your sharing your knowledge.

Hello! I’ve decided now is the time to jump in to sim racing for me, and I’ve already decided on PS5/PSVR2 with a Playseat Trophy and GT7. I’m a techie, an EE, an avid driver, but not really a gamer. This will be just for racing sim.
I just need to decide on the wheel and pedals. I really dig the F1 open style steering wheels’ look and vibe over round. After reading a ton on here and watching reviews, here’s what I’m considering, please help me decide:

1: T300 base, TM Open wheel and LCM pedals. $550 used from Amazon.

Pros: cheaper, f1 wheel, decent pedals Cons: not direct drive, no trueforce, less tactile feedback in GT7

2: Logitech G Pro wheel and pedals. $1298 bundle on Amazon

Pros: everyone says w trueforce it’s the best FFB for GT7 and is thus the most immersive in VR. Much more power for FFB Cons: expensive, no f1 wheel, pedals aren’t great for price
So it’s about $2000 all in with Thrustmaster, and $2750 if I go G Pro. Pretty big price difference.
I have a great “drum simulator”, a custom setup Roland V-drums with lots of drums/cymbals and a 1500w Buttkicker Concert bolted to the drum throne powered by a 3000w Behringer amp. It’s so immersive and real, I’ve been searching for another sim as immersive. GT7/VR2 is the first thing I’ve found that fits the bill and I can’t wait.
Aside from which wheel, I’m trying to understand: is there any benefit to having hand brakes or shifters when only using VR? Or its it best to keep everything on the wheel for max immersion? I used to drive a stick so I’m down, but it seems like grabbing for things that aren’t rendered in VR could take you out.
Should I care about drifting and try to build to it (hand brake/round wheel)? Or is it fine to do it all wheel-only? This is all new to me, I’ve never even driven a sim, but I’ve read enough, I’m sold. I’m just totally inexperienced so thanks for your help.

I have been doing a deep dive into studying and applying TensorRT acceleration to ControlNet for realtime, interactive animation in Stable Diffusion. I have already integrated TensorRT accelerated Stream Diffusion as an img2img pipeline in a realtime-controllable VJ app that uses realtime GANs to generate the driving video, as seen here: https://www.instagram.com/reel/C4AJddYRwdH/?igsh=N2JsejE4dTc0MGhu
I am working on modifying the Stream Diffusion code to add ControlNet -- I have already gotten ControlNet working using Xformers as the accelerator instead of TensorRT and by passing the 12-length down_block_res_samples tuple of tensors and mid_block_res_sample tensor all the way down to the UNet2DConditionModel's forward pass by the unet_step in pipeline.py of Stream Diffusion. This runs at about 12FPS which is kind of.. meh, so I am still working on an adaption of TensorRT accelerated ControlNet.
The progress for where I am at on TRT ControlNet for Stream Diffusion can be found here: https://github.com/cumulo-autumn/StreamDiffusion/issues/132
Note: I am not using the preprocessor, I am piping in already processed frames of openpose skeletons.
Also, my first failed attempt at TensorRT integration for controlnet is not a part of that issues thread, but the details:
I tried to just set up the dynamic inputs and all other input methods in Stream Diffusions tensorrt engine and model code as well as other changes needed to facilitate passing the 12 down_block_res_samples and mid_block_res_sample tensors like initializing space for them in the input buffer after adding them correctly by their reported shapes/sizes to all methods/dicts in any file that had inputs already configured for passing img2img data to the existing img2img unet classes used for tensorrt acceleration. That isn't working due to the graph optimizer still claiming that the input names are invalid for those additional controlnet tensors, even though they are configured in both the get_input_names and get_input_profile methods and as dynamic axes and sample_inputs. I think it has something to do with the graph optimization onnx or other onnx prematurely saving itself to file prior to having the rest of the inputs configured, then the build tries to load the model it just saved which promptly complains about the input names in the submitted profile being invalid. I tried to manually shoehorn them in there in right before the graph from the model is saved but that just got really weird as now it sees all but 2 of the down_sample tensor inputs, and the two it is missing are like number 7 and 12, so it's not like the end of the buffer or anything that makes sense.
That's not a huge deal. It may be possible to get that appraoch working but it's a hack as it's not actually accelerating ControlNet it's just seperating it from the existing TRT acceleration and piping the usable precalculated controlnet tensors for a controlnet image to a Unet retrofitted to accept them. I half expected this to fail as I was trying to be lazy and just see if I could use the already working TRT accelerated unet engine from Stream Diffusion in it's callback to .infer()
I am abandoning this approach and taking the longer, more proper method of a implementing a full Multi ControlNet TensorRT engine implementation. I am making progress on that and should have it hopefully working soon, using Nvidia's SDWebUI plugin implementation of ControlNet as an undocumented guide for this (controlnet/controlnetv2 branches of their TensorRT SDWebUI plugin here: https://github.com/NVIDIA/Stable-Diffusion-WebUI-TensorRT/tree/controlnet_v2 ).
I plan to use this to modify the application I shared at the top of this post, to include a couple of simple but playable 2 player video games using an already working Panda3D driver for controlnet and generic gamepad support, with Panda3D rendering openpose skeleton animations to 3D frames in the background based on the controller inputs and controlnet handling the rest.
As I finish up acceleration of ControlNet, I wanted to bring up AnimateDiff. AnimateDiffV3 has the ability to split up the generation and stitch it seamlessly, and s9roll7's AnimateDiff-CLI fork with LCM variant models can generate animations at a speed of about 3 to 6 frames per second, and this is with multiple controlnets and LoRAs applied (found here https://github.com/s9roll7/animatediff-cli-prompt-travel)
The challenge with AnimateDiff in realtime is likely not even TensorRT acceleration, even though that may indeed be extremely difficult. I haven't looked into it yet, maybe I am lucky and some absolute madlad already made a TRT engine for AnimateDiff?
Anyway, the challenge I am thinking is present with making AnimateDiff not just "realtime" but responsive to realtime input, has been the fact that it renders an entire buffer of frames all at once for an animation, and even then, only V3 can even split an animation into sections like that iirc. So I am not sure if I can split the buffer up into small enough chunks in AnimateDiffV3 to have it responding in realtime to live controller inputs via controlnet.

My two initial questions on tackling realtime AnimateDiff:

1) How small can I make a buffer of frames for each generated segment of AnimateDiff V3 before it gets incoherent/inconsistent between generations? I am assuming you have to have at least a full animation keyframe inside of one of those buffers being generated, is that correct?
and
2) Is there any way for me to split up the generation process in AnimateDiffV3 while it is generating, to apply a buffer of controlnet inputs across the whole or partial set of frames that it is actively working on?
Any insight on this is valuable to me. I am thinking the second question I ask here on architecting a realtime-controllable AnimateDiff TensorRT engine is more viable than just reducing the size of the number of frames beinf generated but it may either be really really hard to figure out or impossible.
I don't care if it's hard to get working, I just want to know if that or something like that is possible

2.33.0.161 -> 2.34.0.19 Version Gszabi Part 2

Rest of the 2.34.0.19 ground vehicle changes by gszabi (continued) (there can be overlaps between this and Oshida's 2.34.0.19 ground vehicle changes, in cases where I clarified or specified something Oshida said, however I do not repeat everything he said, so when looking at a specific vehicle, please look at both parts at once):

• ZSU-23-4, ZSU-23-4 [ITA], ZSU-23-4 [ISR]:
  • Hull LFP at the sides – 9.4 → 14 - 20.4 mm
  • Hull rear – 9.4 → 7.1 mm
  • Holder of the log in the rear – 8 → 2.8 mm
  • Hull floor – 9.4 → 4.8 mm
  • Grille above the engine: 4 → 10 mm
  • Hull roof around the grille – 9.4 → 5.1 mm
  • Hull roof in the rear – 9.4 → 15.2 mm
  • Horizontal armour above the tracks – 10 → 4.1 mm
  • Turret roof in the front – 8.9 → 2.8 mm
  • Turret roof in the rear – 8.9 → 5.3 mm
  • Turret roof at the edges – 8.9 → 7.1 mm
  • Vertical armour around the guns – 8.9 → 3.8 mm
  • Turret ring – 9.4 → 5 mm
  • Turret underside – 8.9 → 4.1 mm
  • Breech – 5 → 150 mm
• ZSU-57-2, ZSU-57-2 [ITA], Ps. 461 (ZSU-57-2), ZSU-57-2 [ISR]:
  • Grille above the engine: 5 → 10 mm
  • Breech – 5 → 150 mm
  • Bug which prevented damage to the breech if it wasn't penetrated, fixed
• 3-inch Gun Carrier:
  • Superstructure inner side – 76.2 → 38.1 mm
  • Superstructure outer side – 25.4 → 38.1 mm
• Centurion Mk. 5 – add-on hull UFP: 50.8 → 44 mm
• Centurion Mk. 10, strv 101, Sho't Kal Gimel – add-on hull UFP: 51 → 44 mm
• Challenger 1 Mk. 3 – hull ERA holder:
  • 10 → 25 mm
  • Volumetric
  • Secondary shrapnel generating disabled
  • Added two more parts, 8 and 25 mm
• Challenger 2 (Dorchester 2E):
  • Commander's and loader's hatches – 44 → 20 mm
  • Add-on armour on the turret rear – 20 mm RHA → 5 mm aluminium
  • Add-on armour on the turret roof rear – 25 mm RHA → 5 mm aluminium
  • Added shell ejection animation
• Challenger 2 TES (Dorchester 2H):
  • Lower hull side composite screens:
    • 250 → 40 mm
    • Added another layer with same stats though
  • Hull side armour the composites are mounted to – composite screen → aluminium alloy 7020
• Challenger 2 TES (Dorchester 2H) "OES":
  • Lower hull side composite screens:
    • 250 → 40 mm
    • Added another layer with same stats though
  • Hull side armour the composites are mounted to – composite screen → aluminium alloy 7020
  • 7.62 mm – ammo: 2300 → 2000 shots
  • 12.7 mm – operator: commander → loader
• Challenger 3 TD:
  • Deadzone in the rear narrowed
  • Minimum elevation in the rear – 3° → 1.5°
• Chieftain Marksman:
  • Add-on armour on either sides of the driver's optic:
    • Volumetric
    • It can now be hidden in the armour viewer
    • It's now shown in X-Ray
• Conqueror Mk. 2:
  • Armour above the breech – CHA → RHA
  • Driver's hatch – CHA → RHA
• Crusader Mk. III:
  • Mantlet – 30 → 88.9 mm
  • Minimum elevation to diagonally front left – -7.5° → -12°
  • Minimum elevation to the left – -7.5° → -8°
• Daimler Mk. II – added 14 mm armour on the turret front below and around the gun
• Mk. VIa A.A. Mk. I – ammo can now be damaged
• Stormer (HVM) – shooting the turret drive can no longer overpressure the vehicle
• TTD:
  • Firewall composites are now red in X-Ray
  • Composite screens above and below the breech no longer can generate secondary shrapnel
• Warrior – 30 mm ammo: 222 → 249 shots
• ZA 35:
  • Parked radar now faces downwards
  • Suspension physics changed
• ZT3 Mk. 3
  • Ammo – 12 → 15 missiles
  • Shooting the loaded ATGM can no longer be fatal
  • Suspension physics changed
• 3½t Truck (SAM-1) – added five 5 mm structural steel boxes
• Chi-Ha, Chi-Ha Kai, Chi-Ha Kai (Twitch Drop), Chi-He, Chi-He (No. 5 S.Ren.), Chi-Nu, Chi-Nu (Experimental 7,5cm), Chi-Ri, Chi-To (Prototype 2), Ho-I, Ho-Ni I, Ho-Ni III, I-Gō Kō (late), Ke-Ni, Long 12cm SPG, Ro-Gō, Short 12cm SPG, So-Ki, STA-3, STB-1, Ta-Se, Type 61 TK, Chi-Ha [CHN], Chi-Ha Kai [CHN] – grille above the engine: 4 mm structural steel → 10 mm RHA
• Experimental 12cm SPG:
  • Grille above the engine – 4 mm structural steel → 10 mm RHA
  • Shooting now depletes the ammo rack
• Ha-Gō, Ha-Gō Commander, Ha-Gō [CHN] – grille above the engine: 4 mm structural steel → 6 mm RHA
• Ho-Ro:
  • Grille above the engine – 4 mm structural steel → 10 mm RHA
  • Vertical limits – -10° / 20° → -5° / 30°
• Ka-Chi – vertical limits without the pontoons: -10° / 20° → -15° / 20°
• Ka-Mi – grille above the engine: 5 mm structural steel → 6 mm RHA
• Na-To:
  • Grille above the engine – 4 mm structural steel → 10 mm RHA
  • Vertical limits – -8° / 19° → -10° / 20°
• TKX No. 3 (2006), TKX No. 3 (2009), Type 10 TK:
  • Grille on the hull rear – 15 → 10 mm
  • Firewall composites are now red in X-Ray
• Type 60 APC (ATM-1) – grille above the engine: 4 mm structural steel → 8 mm RHA
• Type 60 SP (C) – grille above the engine: 4 mm structural steel → 9 mm RHA
• Type 74 TK (C), Type 74 TK (E), Type 74 TK (F), Type 74 TK (G), Type 75 MSSR, Type 87 AW – grille above the engine: 5 mm structural steel → 10 mm RHA
• Type 90 TK:
  • Grille above the engine – 5 → 10 mm
  • Firewall composites are now red in X-Ray
  • Added two 5 mm structural steel boxes
• Type 90 TK (B):
  • Grille above the engine – 5 → 10 mm
  • Firewall composites are now red in X-Ray
• Type 99 HSP:
  • Camera position changed
  • Contrary to what Oshida said, the gearbox didn't change
• 211, ZTZ59A – 7.62 mm: SGMT → QJT59
• CCKW 353 (M45C) – added 3 mm fenders
• PGZ80:
  • Breech – 5 → 150 mm
  • Bug which prevented damage to the breech if it wasn't penetrated, fixed
• VT4A1:
  • Added two 4 mm structural steel boxes in turret rear
  • Driver – 1200x800 NVD → 800x600 thermal
• ZSD63 (PG87) – added two 10 mm rubber-fabric rear mudguards
• ZTZ96, ZTZ96A, ZTZ96A (P) – elevation speed: 5 → 6 °/s
• ZTZ99 (Phase 1), ZTZ99 (Phase 2) – vertical limits: -5° / 13° → -6° / 14°
• ZTZ99A – added two add-on track armour on the hull LFP
• 90/50 M36B1 – exhaust can now be hidden in the armour viewer
• 100/17 3Ro – the front axle is now also driven
• AB. 41:
  • Front hull MG mantlet – 18 → 8 mm
  • Small vertical armour behind each wheel – 8 → 6 mm
  • Armour around the gun barrel – 8 → 30 mm
  • "Frame" around the mantlet – 8 → 20 mm
  • Armour at the base of the turret rear – 6 mm RHA → 50 mm structural steel
  • Added 6 mm grille above the engine
  • Rear hull MG mantlet – 18 → 30 mm
  • Added a 14 mm armour around the driver's viewport
  • Added a 14 mm armour around the machine gunner's viewport
  • Collar behind the turret – 8 mm RHA → 5 mm structural steel
  • Fenders – 8 mm RHA → 3 mm structural steel
• Ariete (AMV PT 1), Ariete (P), Ariete (PSO), Ariete (WAR) – engine sound: Leopard 2 A5 → Ariete
• Centauro MGS – turret sides are now volumetric
• KF41:
  • Added an 8 mm HHRHA armour on top of the missile launcher
  • Missile ammo – 2 → 4 missiles
• OF 40 Mk. 2, OF 40 Mk. 2 (MTCA) – 12.7 mm ammo: 1200 → 400 shots
• Turán I, Turán II, Turán III:
  • Hull MG removed from X-Ray
  • Roadwheels on the hull rear – 15 → 10 mm
• VBM Freccia C/C:
  • Camera position changed
  • Shooting the loaded ATGM can no longer be fatal
• VCC Dardo:
  • 22 mm mantlet part can no longer be hidden in the armour viewer
  • 5 mm small rectangles on the lower corners of the hull UFP can no longer be hidden in the armour viewer
• AMC 34 (APX 1) – vertical limits: -8° / 12° → -18° / 20°
• AMX 10 P – vertical limits: -5° / 50° → -8° / 50°
• AMX 30 DCA:
  • Engine sound – M4A1 → AMX 30
  • Boxes above the tracks – RHA → structural steel
• AMX 40:
  • Firewall composites are now red in X-Ray
  • Vertical limits – -7° / 20° → -8° / 20°
• AMX 50 Foch – 15 mm mg – vertical limits: -3° / 20° → -12° / 12°
• AMX 50 Surbaissé – turret floor is now 50 mm
• AMX 50 Surblindé:
  • Mantlet – 85 → 105 mm
  • Turret lower front – 100 → 156 mm
  • Turret floor is now 50 mm
  • 20 mm – vertical limits:
    • Diagonally front left – -17° / 63° → -13° / 37°
    • From there to diagonally rear right – -17° / 63° → -17° / 37°
    • From there for 7 more degrees to the rear – -12° / 63° → -12° / 37°
    • From there for 10 more degrees to the rear – -9° / 63° → -9° / 37°
• H 35:
  • Turret front – 45 → 40 mm
  • Turret rear – 40 → 47 mm
  • Commander's cupola – 40 → 25 mm
  • Turret roof in the front – 25 → 40 mm
  • Armour in front of the commander's optic – 35 → 25 mm
  • Armour in front of the turret side optics – 35 → 25 mm
  • 25 mm armour around the gun split into 24 mm lower and 14 mm upper halves
• Leclerc s. 1, Leclerc s. 2, Leclerc s. 2 (AZUR), Leclerc s. 21:
  • Firewall composites are now red in X-Ray
  • They now use unique Leclerc track sounds
• P.7.T. AA:
  • Steering deflection – 47° → 41°
  • Suspension physics changed
• R 35 (SA 38):
  • Hull floor – 5 → 10 mm
  • Hull LFP – 32 → 43 mm
  • Hull lower UFP – 25 → 30 mm
  • Middle and right part of hull lower UFP – 25 → 21 mm
  • Hull upper UFP – 32 → 40 mm
  • Hull rear – 40 → 32 mm
  • Hull rear hatches – 40 → 32 mm
  • Hull roof – 25 → 14 mm
• K9 (VIDAR):
  • Added a 35 mm internal armour in front of the driver
  • Added a 35 mm armour around the gunner's optic
  • Added a 20 mm internal armour at the breech
• pbv 501 – chassis: tracks → chassis structural steel
• Ps. 273 (Leopard 2 A4 FIN):
  • Firewall composites are now red in X-Ray
  • Hull side composite screens – 100 → 110 mm
  • Side mudguards – 5 → 12 mm
• Ps. 274 (Leopard 2 A6 NL):
  • Firewall composites are now red in X-Ray
  • Armour at the driver's optics – 30 → 45 mm
  • Hull side composite screens – 100 → 110 mm
  • 7.92 mm deadzone around the commander's optic made larger
• strv 122A, strv 122B+:
  • Firewall composites are now red in X-Ray
  • Armour at the driver's optics – 30 → 45 mm
• strv 122B (PLSS):
  • Firewall composites are now red in X-Ray
  • Armour at the driver's optics – 30 → 45 mm
  • 7.62 mm – vertical limits
    • To the left – -10° / 50° → -10° / 20°
    • Diagonally to rear right – -8° / 50° → -8° / 75°
• strv m/39, strv m/39 (Twitch Drop), strv m/40L – vertical limits: -15° / 25° → -14° / 20°
• M-51 (M4A3), M-51 Gimel – roadwheels above the left track: 15 → 10 mm
• Merkava Mk. 2 Bet [USA], Merkava Mk. 1 Bet, Merkava Mk. 2 Bet, Merkava Mk. 2 Dalet – added a 5 mm armour inside the exhaust
• Merkava Mk. 4 Bet, Merkava Mk. 4 LIC, Merkava Mk. 4 Meil Ruach:
  • Firewall composites are now red in X-Ray
  • Grille above the engine – 5 → 10 mm
  • Added a 5 mm armour inside the exhaust
• Sho't Kal Dalet, Sho't Kal Dalet [UK]:
  • Add-on hull UFP – 51 → 44 mm
  • Wheel on the ram in the front – 20 → 15 mm

Rest of the 2.34.0.19 naval vehicle changes by gszabi (there can be overlaps between this and Oshida's 2.34.0.19 naval vehicle changes, in cases where I clarified or specified something Oshida said, however I do not repeat everything he said, so when looking at a specific vehicle, please look at both parts at once):

• Alaska (CB-1), Allen M. Sumner (DD-692), Arizona (BB-39), Arkansas (BB-33), Atlanta (CL-51), Baltimore (CA-68), Bennion (DD-662), Cleveland (CL-55), Des Moines (CA-134), Detroit (CL-8), Fargo (CL-106), Fletcher (DD-445), Helena (CL-50), Nevada (BB-36), New Orleans (CA-32), Newport News (CA-148), Northampton (CA-26), North Dakota (BB-29), Pensacola (CA-24), Pittsburgh (CA-72), Portland (CA-33), Raleigh (CL-7), Texas (BB-35), Trenton (CL-11), Wyoming (BB-32), Admiral Graf Spee, Admiral Hipper, Helgoland, Kaiser, Karlsruhe, Köln, Leipzig, Nassau, Nürnberg, Ostfriesland, Prinz Eugen, Sachsen, Scharnhorst, Westfalen, Von der Tann, Z 15, Z 35, Z 46, Chapayev, Frunze, Imperatritsa Mariya, Kerch, Kerch, Kirov, Krasny Kavkaz, Krasny Krym, Kronshtadt, Leningrad, Marat, Maxim Gorky, Mikhail Kutuzov, Neustrashimy, Novorossiysk, Opytny, Parizhskaya Kommuna, Poltava, Shcherbakov, Sverdlov, Tashkent, Voroshilov, Zheleznyakov, Arethusa (26), Belfast (C 35), Colossus, Daring (D 05), Dido (37), Dreadnought, Enterprise (D 52), Glorious (77), Hawkins (D 86), Hood (51), Invincible (85), Kent (54), Leander (75), Liverpool (11), London (69), Norfolk (78), Orion (52), Renown (72), Southampton (83), Tiger (C 20), Tobruk (D 37), Valhalla (F 9A), York (90), Agano, Aoba, Furutaka, Fusō, Haruna, Hatsuzuki, Hyūga, Ise, Isuzu, Kongō, Mikuma, Mogami, Myōkō, Sendai, Settsu, Shimakaze, Suzuya, Yamashiro, Yūdachi, Andrea Doria, Attilio Regolo, Bartolomeo Colleoni, Conte di Cavour, Dante Alighieri, Duilio, Etna, Eugenio di Savoia, Geniere (D 555), Impetuoso (D 558), Leonardo da Vinci, Leone (LE), Luigi di Savoia Duca degli Abruzzi, Pola, Trento, Zara, Colbert, Courbet, Duguay-Trouin, Dupleix, Émile Bertin, Jeanne d'Arc, Primauguet
  • Module HP reorganised
  • Crew reorganised
• Asheville (PGM-84), Douglas (PGM-100), Kim Qui (HQ 605), PC-451, PC-466, Pr. 122A, Pr. 122bis (MPK-163), Pr. 122bis (MPK-173), Grey Fox (SGB 4), Grey Goose (SGB 9), Ka-Rō Model 2, No. 1, No. 4, No. 13, No. 14, Freccia (P 493), Saetta (P 494), Stefano Turr:
  • Module HP reorganised
  • Engine, transmission:
    • Shrapnel damage multiplier – 1.5 → 5
    • Secondary shrapnel damage multiplier – 0.05 → 1
  • Funnel – armour type: engine → structural steel
• Aylwin (DD-355):
  • Module HP reorganised
  • Center of gravity shifted
• Barker (DD-213):
  • Module HP reorganised
  • 20 mm AA turrets:
    • Narrowed deadzone forwards and backwards
    • Minimum elevation sideways – -5° → -6°
• Brooklyn (CL-40), Cowell (DD-547), Davis (DD-395), Frank Knox (DD-742), Gearing (DD-710), Litchfield (DD-336), Mitscher (DL-2), Moffett (DD-362), Phelps (DD-360), Porter (DD-356), Somers (DD-381), Welborn C. Wood (DD-195), Elbing, Emden, Jaguar, K 2, Leopard, Luchs, T 22, Z 5, Z 20 (1941), Z 20 (1945), Z 25, Z 33, Z 47, Besposhchadny, Bezuprechny, Blagorodny, Bravy, Moskva, Ognevoy, Ryany, Spokoiny, Smelyi, Soobrazitelny, Stroyny, Abdiel (M 39), Brissenden (L 79), Cadiz (D 79), Calpe (L 71), Churchill (I 45), Diana (D 126), Eskimo (F 75), Garland (H 37), Grafton (H 89), Haida (G 63), Iron Duke (18), Jervis (G 00), Kelvin (F 37), Marlborough (79), Matapan (I 43), Mohawk (F 31), Nepal (G 25), Akizuki, Harukaze (DD-101), Hatsuharu, Ikoma, Kako, Kuma, Momi, Nenohi, Tone, Yūbari, Yūgure (DD-184), Aquila (AQ), Aquilone (AL), Aviere (AV), Camicia Nera (CN), Comandante Margottini (MA), Geniere (GE), Giovanni da Verazzano, Raimondo Montecuccoli, Tigre (TI), Turbine (TB), Aigle (X 13), Bourrasque (T 41), Chacal (X 23), La Combattante (L 19), Le Malin (X 82), Léopard (X 22), Lorraine, Marceau (D 601), Milan (X 111), Panthère (X 43), Vauquelin (X 53), Vautour (X 71): Module HP reorganised
• LCM(6)(F), BMO-318, Pr. 161 (1943), Pr. 161 (1944), Pr. 191, Pr. 191M, LCS (L) 257, No. 1 "Sakigake":
  • Module HP reorganised
  • Engine, transmission – KE damage multiplier: 1.5 → 5
• LCS(L)(3)-1, SC-497, Tucumcari (PGH-2), Albatros (S 61), Artur Becker (841), Hugin (P 6191), Jaguar (S 1), MZ 1, Pfeil (P 6193), Pr. 201, Pr. 201M, Pr. 206, Pr. 206M, Pr. 1204 (AK-222), Pr. 1204 (AK-508), Pr. 12412, Pr. 12412P, Asagao (YTE-01), Ka-Rō Model 4, Ke-I Kō, PG No. 2 (PG-822), PT No. 8 (PT-808), PT No. 15 (PT-815), Se-Ha, Folgore (P 490), Sparviero (P 420):
  • Module HP reorganised
  • Engine, transmission:
    • Shrapnel damage multiplier – 1.5 → 5
    • Secondary shrapnel damage multiplier – 0.05 → 1
• Mississippi (BB-41):
  • Module HP reorganised
  • Crew reorganised
  • Quad 40 mm AA turret above the bridge – minimum elevation: -13° → 10°
  • Two quad 40 mm AA turrets to either side of the funnel – minimum elevation: -12° → 10°
• PT-3, PT-20, PT-59, PT-71, PT-103, PT-109, PT-174, PT-314, PT-556, PT-565, PT-625, PT-658, PT-810, PT-811, PT-812, PTF-7, Krischan der Große, KS 5, LS 3, LS 4, Nr. 401, R 41, R 130, S 100 (1944), D-3, G-5 (TK No. 106), MO-4, OD-200, Pr. 123bis, Pr. 123K, Pr. 123K (A-10bis), Pr. 183, Ya-5M, Dark Adventurer (P 1101), Dark Aggressor (P 1102), Dark Aggressor (P 1102) (Twitch Drop), Gay Archer (P 1041), MGB 61, MGB 75, MGB 312, MGB 332, ML 1383, MTB 1, MTB 2, MTB 379, MTB 396, MTB 422, No. 1 (T-1), No. 538 (T-14), No. 538 (T-14) (Type 96), MS 15, MS 53, MAS 551, MAS 561, MAS 569, VAS 201:
  • Module HP reorganised
  • Engine, transmission – 5 → 2 mm
• PT-6, R 301, S 38, S 62, S 100 (1945), S 204, S 701, VS 10, Pr. 183 (A-215), Brave Borderer (P 1011), No. 10 (T-51a), No. 11 (T-51b), MC 485, MS 461, MS 472, MV 473, MV 611:
  • Module HP reorganised
  • Crew reorganised
  • Engine, transmission – 5 → 2 mm
• Roanoke (CL-145):
  • Module HP reorganised
  • Crew reorganised
  • Armour to the sides of the last two primary turrets' ammo storage – RCA → RHA
  • Center of gravity shifted
• AF 99, SF 05, SF 07:
  • Module HP reorganised
  • Engine, transmission – 5 → 4 mm
• AF 128:
  • Module HP reorganised
  • Crew reorganised
  • Engine, transmission – 5 → 4 mm
• Bayern:
  • Module HP reorganised
  • Crew reorganised
  • Camera position changed
• Köln (F 220), Lübeck (F 224):
  • Crew reorganised
  • Added two more engine rooms
• VS 8 Schell I:
  • Module HP reorganised
  • Crew reorganised
  • Engine, transmission – 5 → 2 mm
  • Nose 15 mm turret: minimum elevation backwards – 5° → 19°
• G-5, G-5 (ShVAK-20):
  • Module HP reorganised
  • Engine, transmission – 5 → 2 mm
  • 12.7 mm / 20 mm turret – minimum elevation forwards: 42° → -5°
• Letuchiy:
  • Module HP reorganised
  • Torpedo ammo not being spent in X-Ray fixed
  • Added a duplicate HE
• Pr. 186, Pr. 186 (MK 85), Pr. 1124 (1942), Pr. 1124 (1944), Pr. 1124 (1945), No. 36:
  • Module HP reorganised
  • Crew reorganised
  • Engine, transmission – KE damage multiplier: 1.5 → 5
• Arrow (P 88), MTB 5001, No. 241 (T-38):
  • Module HP reorganised
  • Crew reorganised
  • Engine, transmission – 5 → 2 mm
  • Funnel – armour type: engine → structural steel
• MGB 601, MTB 617, MTB 697, ML 100, ML 345:
  • Module HP reorganised
  • Engine, transmission – 5 → 2 mm
  • Funnel – armour type: engine → structural steel
• Kurama:
  • Module HP reorganised
  • Lowest part of the bridge elevator – 50.8 mm RHA → 101.6 mm RCA
  • Horizontal armour at the bow and stern – 38.1 → 50.8 mm
  • Roof of the bridge – 50.8 → 76.2 mm
  • New 50.8 mm armour around the 200 mm turrets' elevators
  • Belt armour at the bow and stern – 102 → 101.6 mm
  • 305 mm turrets:
    • Front and back – 229 → 228.6 mm
    • Sides – 152 → 152.4 mm
  • 200 mm turret floors – 38.1 → 7.62 mm
  • 47 mm turret fronts – RCA → RHA
  • 177.8 mm internal gun shield removed from the first 305 mm turret
• Shōnan – single 120 mm turret → double 120 mm turret
• Le Triomphant (X 83):
  • Module HP reorganised
  • 20 mm turret amidships – horizontal limits: -125° / 155° → -180° / 180°
  • 20 mm turret to the right of the second primary turret – horizontal limits: -60° / 180° → -180° / 45°
• Paris:
  • Module HP reorganised
  • Crew reorganised
  • Added two more primary shell storages and one more secondary ammo storage
  • Removed six fuel tanks
  • Added eleven coal bunkers
  • Lowest part of the bridge elevator – 50 → 60 mm
  • Small belt part at the bow – 250 → 210 mm

Rest of the 2.34.0.19 aircraft gun changes by gszabi (there can be overlaps between this and Oshida's 2.34.0.19 aircraft gun changes, in cases where I clarified or specified something Oshida said, however I do not repeat everything he said, so when looking at a specific weapon, please look at both parts at once):

• 7.7 mm Type 1 – API-T now has tracer
• 13 mm M.G. 131 – HEI now has tracer
• 23 mm Model 2K – API-T penetration: 34.41 → 31.29 mm
• 30 mm GIAT M781 – SAPHEI:
  • Speed – 775 → 810 m/s
  • Penetration – 27.15 → 28.92 mm
• 30 mm Ho-155:
  • Stock belt – HE / AP-T / HEFI / HEF-T → HE / AP-T / HEFI
  • Universal belt – HEF-T / HEFI / AP-T / AP-T / HE / HE → HEFI / AP-T / AP-T / HE / HE
  • Ground targets belt – HEF-T / AP-T / AP-T / AP-T → HEFI / AP-T / AP-T / AP-T / AP-T
  • Tracers belt – HEF-T → AP-T
  • Stealth belt – HE / HE / HE → HE
  • AP-T:
    • Mass – 378 → 235 g
    • Speed – 770 → 700 m/s
    • Penetration – 44.96 → 27.99 mm
  • HE, HEFI:
    • Mass – 350 → 235 g
    • Speed – 770 → 700 m/s

Current Dev version : 2.34.0.28
Current Dev-Stable version : 2.33.0.168
Current Live version : 2.33.0.168