How the conversion works

The recovery tool reads your old crosshair settings, carries over the ones that have a known equivalent in the new system, and asks you to choose the rest. This page explains each rule, the evidence behind it and what has not been measured yet. For what changed in the game itself, see the CS2 crosshair update.

The rules are fixed code, so the same input always produces the same proposal. The current rule set is assisted-2026-09-23.1.

What carries over, setting by setting

The labels in the last column are the ones you may see next to each value in the tool. Any value you change yourself is labelled "Chosen by you".

Color preset names: Red comes from community references only; the other preset names come from the old game files.

Old styles

Old style numbers meant slightly different things in CS:GO and in CS2 before the update. Whether you keep the starting answer, Pre-update CS2, or answer Not sure, the tool treats them as follows. Where you must choose, the tool may offer likely styles, but it never picks one for you.

Any other old number is not recognised. Your other settings are kept and you choose a new style.

The site owner tested one suggested workaround for Classic, Split Size Ratio 0 with Inner Split Alpha 1, and it did not widen the crosshair. It is not offered as a migration rule.

Old settings with no new equivalent

These old settings were removed by the update and have no replacement in the new menu. The tool keeps your old value visible and lists it in the settings checklist, so nothing disappears silently.

• Outline thickness. The new Outline control is only on or off.
• Use weapon gap. Your old gap may have changed with each weapon, so one old gap number may not describe what you saw.
• Fixed crosshair gap. There is no new setting for it.
• Color preset. The new menu has no presets; see the Color row above.
• Use transparency (on/off). The new alpha value alone sets opacity.

Why there is no single multiplier

Valve published no formula, and the measurements show why one multiplier cannot work:

• The old game drew size and thickness at 2.25 pixels per unit at 1920×1080 and 2 (the captures allow about 1.9643 to 2.0313) at 1280×960 (4:3), rounded, with thickness at least one pixel. That is the render height divided by 480; old-game captures at 1600×1200 (2.5 pixels per unit) fit it too, but no conversion is offered there because the new game was not captured at that resolution.
• The old gap was not scaled: the space next to the crossing bar was 4 + your gap setting in pixels, with fractions dropped.
• The new game draws Length and Thickness in pixels and counts Gap from the centre of the crosshair, not from the crossing bar.
• So the new Gap depends on the new Thickness, and any multiplier would be wrong for most gaps. The old sizes also scaled with the screen height, so the numbers for other resolutions are expected to differ and are not guessed.

Outside the measured setups, Length, Thickness and Gap start unresolved. You type your own values or choose Start with sample sizes, which fills Length 8, Thickness 3 and Gap 1. That sample is a starting point you chose. It is not a conversion of your old sizes and not the game default (Length 8, Thickness 2, Gap 4).

How each value is labelled

Every new value records two separate things: where it came from, and what happened to the old value on the way. Keeping them apart stops a value you chose from being presented as a conversion. In the tool, a value you or the editor sample supplied shows its origin label ("Chosen by you" or "Editor sample"); every other value shows its outcome label. Old settings the update removed appear in the side-by-side table as "No new equivalent".

Where a value came from

What happened to the old value

Evidence categories

Every rule, range and command on this site cites evidence. Each piece of evidence belongs to one category:

Evidence list

Links to game files point to fixed versions of the public GameTracking-CS2 mirror. Screenshots and console sessions from the site owner are listed but not published.

Command coverage

Each new console command has one of these statuses:

• Verified: the site owner set or selected the value in the game and read it back in the console.
• Documented: the command, its type and its range appear in the public console variable list for the current build, and the game menu binds the matching control to it.

The tool only outputs commands with one of these two statuses. Reading back the whole documented set in game is still pending, so treat the game as the final check. Right now 4 settings have verified commands and 11 have documented ones.

Command names come from a fixed list and values are checked against each command's range, so nothing you type can become part of a command. The old commands below are never included, because the update removed or hid them:

cl_crosshairsize, cl_crosshairthickness, cl_crosshairgap, cl_crosshairalpha, cl_crosshairusealpha, cl_crosshaircolor, cl_crosshair_outlinethickness, cl_fixedcrosshairgap, cl_crosshairgap_useweaponvalue

The longest line the game console accepts has not been checked. The tool therefore also splits a long command line into blocks of at most 120 characters, and offers the commands as a crosshair.cfg file.

Versions

Game builds checked

The console checks by the site owner did not record a build number.

The old code decoder

Old crosshair codes are decoded with the byte layout of akiver/csgo-sharecode at commit 753f16f (version 5.0.0), used under the MIT licence, Copyright (c) 2017-present AkiVer. Only its decoding logic is reproduced. Its command generator is not used, because it writes the old command names.

This site adds stricter checks that the reference does not make:

• Input is limited to 256 characters and must have the exact code shape.
• Every character is checked against the code alphabet before any decoding. Letter case is never changed.
• The decoded length, the checksum, the format marker byte, reserved bits and bytes, and the range of each field are all checked.
• Syntax errors, integrity errors and unsupported formats get different messages. A passing checksum alone does not prove a code is a crosshair, because match share codes use the same alphabet and length.
• Codes with the newer format marker are refused instead of being read as old settings.

Split distance: the reference reads split distance as a number from 0 to 7, but the old menu allowed up to 16. This site reads a wider part of the byte and flags any value above 7 as uncertain instead of refusing the code. In every checked fixture the split distance is between 3 and 7, where both readings agree.

The project's fixture file lists 9 old codes checked against the pinned reference; 9 of them decode and re-encode to the identical code. It also lists 3 codes that must not be read as old crosshairs: a new-format code from the updated game and match share codes. This establishes old-code decoding only. It does not show that the updated game imports those codes or that any conversion looks the same.

How sizes are converted

Sizes are converted from measurements, not from a guessed formula. In a capture session the same crosshairs were set in CS2 before the update (still available as a Steam beta) and in the current game, a lossless full-size screenshot was taken of each, and a script measured every bar in pixels.

At each measured setup every crosshair fits these rules exactly. Only the old game's scale depends on the resolution: 2.25 pixels per unit at 1920×1080 and 2 (the captures allow about 1.9643 to 2.0313) at 1280×960 (4:3).

For example, at 1920×1080 the old Classic Static 2.1 / 0.3 / −2 drew bars 5 pixels long and 1 pixel thick with 2 pixels of space, so it becomes Length 5, Thickness 1, Gap 3. At 1280×960 the same crosshair drew bars 4 pixels long, so there it becomes Length 4, Thickness 1, Gap 3. For each setup, thirteen further crosshairs, kept out of the measurement and chosen with awkward fractions, were then set in the new game at the converted values: every one matched its old screenshot pixel for pixel.

The tool fills sizes in only when everything matches what was measured: an old Classic Static crosshair from CS2 before the update, the new Static Cross style, your resolution and values inside the covered range. It asks for the two facts it cannot read from a code: where your crosshair is from (Pre-update CS2 unless you change it) and the resolution you play at. Anywhere else you choose the sizes, and the measured values for 1920×1080 and 1280×960 (4:3) are shown as a guide. The switch for this build is on.

Sources: E-CAPTURE-OLD-1080; E-CAPTURE-NEW-1080; E-CAPTURE-HOLDOUT-1080; E-CAPTURE-OLD-960; E-CAPTURE-NEW-960; E-CAPTURE-HOLDOUT-960; E-CAPTURE-OLD-1600; E-OWNER-BETA-BRANCH.

How a setup is validated

A profile moves through three states:

Before a profile can become validated, the site's calibration plan requires at least:

• Paired captures of the same crosshair in the old and the new game under controlled conditions, recording the builds, the actual render resolution, aspect ratio, display mode (native, letterbox or stretched) and the full unresized image.
• At least five distinct values for each size dimension that is fitted separately.
• At least ten mixed configurations held back from fitting and used only for testing, for each style and resolution class.
• No more than one native pixel of error for each measured part of a static crosshair on those held-back tests.

These are a minimum starting point, not proof that a profile is good enough. A profile only applies to the styles, source versions and value ranges it was measured on, and only for the matching game version. One screen setup does not validate another: a 16:9 result says nothing about stretched 4:3. Dynamic behaviour and Follow Recoil need measurements over time, which static captures cannot provide.

Preview limits

The preview is a schematic, not a copy of the game's renderer. It is labelled "Schematic preview — check the result in CS2". It draws your settings in abstract units. Bars start Gap units from the centre and never reach inside the crossing bar, as measured for Static Cross at 1920×1080 and 1280×960. Thickness 0 draws no bars, as in the measurement.

The preview never draws your old crosshair scaled by a guessed formula next to the new one.

New-format game codes

This site cannot read or create the new game's crosshair codes. Their layout is unknown; only a changed format marker has been inferred. The website link the tool creates is not a game code, and the console commands are not a game code either. The FAQ explains the difference.

Last checked

Evidence last checked on 23 September 2026, against public game data for builds 2000913 and 2000914. Changes to rules and targeted builds are recorded in the changelog. If something here looks wrong, please use the contact page. For help using the tool, see how to use it.