> ## Documentation Index
> Fetch the complete documentation index at: https://docs.phones.farm/llms.txt
> Use this file to discover all available pages before exploring further.

# Common Issues

> Diagnostic guide and troubleshooting checklist for iOS device connectivity, USB synchronizer drops, and coordinate calibration.

## 1.Common Issues

#### 1. Mouse Pointer Moves in Opposite Direction

**Problem Description**:

* Moving mouse right, cursor moves left
* Moving mouse up, cursor moves down

**Cause**: iPhone screen auto-rotation causing coordinate system misalignment.

**Solution:**[Lock screen orientation](https://docs.phones.farm/iphone-farm/iphone-setup.md#4displaybrightness)

## 2. Mouse Movement Distance Inaccurate

**Problem Description**:

* Moving mouse a certain distance, cursor moves too fast or too slow

**Cause**: iPhone AssistiveTouch tracking sensitivity settings are incorrect.

**Solution**: [1.**Tracking Sensitivity -Max**](https://docs.phones.farm/iphone-farm/iphone-setup.md#1enable-assistivetouch)   [**2.Tracking SPeed -Minimum**](https://docs.phones.farm/iphone-farm/iphone-setup.md#3enable-trackpadmouse)\
 After setting, **restart screen mirroring** or **reset mouse** to restore normal operation

## 3.Can Click but Cannot Swipe

**Problem Description**:

* Click operations work normally but cannot perform swipe, drag, and other gesture operations

**Cause**: Touch gesture function is disabled.

**Solution**:

1. Open iPhone **Settings**, go to **Accessibility** → **Touch** → **AssistiveTouch**
2. Scroll down to find **Touch Gestures** and **enable** this option

![](https://830878652-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FqV5ltwNxGXLjqsyNgAgu%2Fuploads%2FpzpKJdNLQewWIdJfuVud%2Fimage.png?alt=media\&token=bdc2f9e9-8329-47c6-bdf5-f71144ba7c61)

## 4.Keyboard Issue

#### 1.Cannot Type

**Problem Description**:

* Keyboard input has no response, keystrokes cannot input to iOS device

**Cause**: iOS Full Keyboard Access feature not enabled.

**Solution**: [Enable Keyboards](https://docs.phones.farm/iphone-farm/iphone-setup.md#2enable-keyboards)

#### 2.Shortcuts Not Working

**Problem Description**:

* Using API to input `CTRL + A` (select all) and other common shortcuts has no effect

**Cause**: iOS uses the **COMMAND key**, not the **CTRL key** commonly used in Windows/Linux.

**Solution**:

Important Note

On iOS devices, you should use the **COMMAND** key instead of the **CTRL** key to execute shortcut operations.

## 5.OCR Text Recognition

iClick provides built-in OCR (Optical Character Recognition) functionality that can meet the text recognition needs of most automation scenarios. This article explains the advantages of built-in OCR and why we don't integrate third-party OCR libraries like PaddleOCR.

#### How It Works

iClick's built-in OCR **performs recognition on the iPhone device**, not on the computer host.

**Core Advantages**:

* ✅ **Multi-language Support**: Based on iOS system capabilities, supports 18+ major languages worldwide without additional configuration
* ✅ **Zero Host Resource Usage**: Recognition process completed on iOS device, does not consume computer CPU, memory, or GPU resources

#### Applicable Scenarios

Built-in OCR can meet most automation needs:

* ✅ App interface text recognition / button and menu text extraction
* ✅ Message notification content recognition / in-game text recognition
* ✅ Simple data collection tasks

Recommendation

If your needs **do not involve large-scale document scanning**, built-in OCR is completely sufficient, no need to consider third-party OCR libraries.

#### Why Not Integrate Third-party OCR (Using Baidu PaddleOCR as Example)

#### 1. Model Files Too Large, Integrating Single Language Makes No Sense

PaddleOCR and other deep learning OCR library model files are **very large**:

* Single language model: typically 100MB - 500MB
* Multi-language support: requires downloading multiple models, easily exceeding **1GB - 3GB**
* High-precision models: can reach **5GB or larger**
* Each language service requires separate model files

#### 2. GPU Version Requirements Are Demanding

**Hardware Limitations**

* ⚠️ **Only supports NVIDIA graphics cards** (AMD cards, integrated graphics cannot be used)
* ⚠️ Requires mid to high-end cards: **RTX 2060 or higher**
* ⚠️ Official **CPU version exists but efficiency makes it impractical**

**Performance Bottleneck**

Even with a dedicated graphics card, performance is very limited. Example with **RTX 3060 + 15 iPhones** for button recognition:

```
Configuration: NVIDIA GeForce RTX 3060 (12GB)
Devices: 15 iPhones
Optimization: Extreme concurrency control and performance optimization
Result:
  - GPU already running at full capacity (100% usage)
  - OCR recognition speed unbearably slow
  - Overall performance severely degraded
```

#### You Can Still Use Third-party OCR

If you still need to use third-party OCR libraries like PaddleOCR (such as for doc scan, specific language recognition, etc.), and you have a **high-performance NVIDIA graphics card**, you can **integrate it yourself**.

#### Advantages of Self-integration

In fact, building PaddleOCR and other OCR libraries into independent services is **very simple**:

* 🚀 **Minimal Code** - Building a basic OCR HTTP service requires only dozens of lines of code
* 🔧 **Flexible Control** - Can choose desired language models, customize features and parameters
* 🎯 **Specialized Optimization** - Optimize models for specific scenarios (such as captchas)

#### Applicable Scenarios

Third-party OCR is suitable for the following special scenarios:

| Scenario | Recommended Solution | Description |
| - | - | - |
| Captcha Recognition | Specialized captcha model | Targeted training, higher recognition rate |
| Specific Languages | Corresponding language model | Such as minority languages, dialects |
| Handwritten Text | Handwriting recognition model | Built-in OCR has limited handwriting support |
| Table Recognition | Table-specific model | Structured data extraction |
| Document Scanning | High-precision model | Bulk document processing |

## 6.Precautions for transferring large quantities of device files

iOS instant file transfer uses local area network wireless transmission. The actual transfer speed depends on the router's performance. With 20+ devices, you can use an RJ45 network cable to increase throughput, or transfer files in batches at intervals.

## 7.Precautions for device landscape mode

Landscape mode is supported by default and no additional settings are required. In actual development, portrait mode uses the same coordinate system. The console allows you to manually set the landscape display mode. If you encounter coordinate reversal upon entering the game, this is an iOS bug. Please unlock the screen lock switch first, then use the iOS dot menu -> Device -> Rotate to the correct orientation. Remember to lock the screen orientation again after setting it.

## 8.How to set up your phone to use a SIM card, or for ESIM to access the network.

To access the internet using cellular data, please refer to the following steps.

1. Open Settings -> Wi-Fi -> Connected Wi-Fi -> Edit
2. IPv4 address -> Click Edit and select Manual Mode
3. Enter the IP address and subnet mask. Do not enter the router/gateway field! Do not enter the router/gateway field! Save and the changes will take effect immediately without affecting screen mirroring. If using a wired connection, the entry point is Settings -> Ethernet.

[YouTube video player](https://www.youtube.com/embed/Vyo96wZ72Gs)

## 9.How to write scripts for debugging on another device

With the opening of cloud phone functionality, the new version of iOS instant casting, for security reasons, treats all IPs except 127.0.0.1 as external links and requires authorization. If you need to use one computer for coding and another computer for mobile debugging, please use Windows forwarding.

```
netsh interface portproxy add v4tov4 listenaddress=0.0.0.0 listenport=External Port connectaddress=127.0.0.1 connectport=23188
```


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.