After consistently downgrading the espidf framework versions, I found a version (3.2.1 according to PlatformIO numbering), where there are no artifacts. But in this version there is a bug - soft UART does not work, ESP32-S2 just hangs. In the next one (3.3.0 according to PlatformIO numbering) and the current one, UART works, but there are artifacts on the display.
Should I move to lvgl/lv_platformio with this question or stay here?

Hello again :)
I'll check this on HW as soon as I get it.

Meanwhile to your other questions:
If you are using esp-idf then ESP32-S2 is supported from v4.2 onwards.
ATM, I'm not sure whether this is PlatformIO related... once I get it working we can debug this further.

I can see from your activity on GitHub that you really want to make LVGL happen in your projects, I'm sorry that it is not in a state that would enable you to use full potential of LVGL (and ESPs).

We are currently working on making this better.
If you're interested, please contact me directly at tomas.rezucha@espressif.com and describe requirements of your project, ie. chip/board you are using + LCDs that you intend to use, or any other relevant requirement. If we could prepare an example that would exactly fit your needs, we could kill two birds with one stone.

Looking forward to hearing from you,
Tomas Rezucha

After I managed to solve the problem with the touchscreen, I decided to try the fresh version LVGL. I started with an example from the archive controlPanel.zip that I took here.
I have successfully migrated the project to PlatformIO and launched it. There are no artifacts.
However, I spent a lot of time synchronizing the coordinates of the touchscreen with the orientation of the display. But this is a separate topic for discussion.
It's time to move on, I thought, and I updated both LVGL and the drivers in the project.
First try.
Setting display backlight without PWM - I get a black screen..
Setting display backlight with PWM - I get a white screen.
And that's all.
Second try.
I returned the drivers from the archive. The example display appeared, but again with artifacts.

I am in complete despair, confusion and upset feelings.

@SinglWolf From the picture it appears the artifacts are static? I think I see some on my display as well. I also see artifacts that appear and disappear. Do you see those as well, or just these static pixels? Perhaps I can help debug on my display to help?

From the picture it appears the artifacts are static? I think I see some on my display as well. I also see artifacts that appear and disappear. Do you see those as well, or just these static pixels? Perhaps I can help debug on my display to help?

Artifacts are static if the object is static, because the wrong pixels are always on the right vertical edge of the object. Show a screenshot of your own display showing the artifacts yourself. I'll say it or not.

After connecting the display with the ILI9341 controller to the ESP32-S2, the artifacts disappeared.
The problem is in the drivers for the ILI9488 and ILI9486 controllers.

I was able to completely get rid of the artifacts on the display screen with the ILI9488 controller using very old drivers.
Here is from my project set of functions which there are no artifacts on the display.:

void disp_spi_add_device(spi_host_device_t host)
{
    gpio_num_t cs;
    gpio_get_lcd(&cs, NULL, NULL);
    ESP_LOGI(TAG, "Adding SPI device");
    ESP_LOGI(TAG, "Clock speed: %dHz, mode: %d, CS pin: %d",
             SPI_TFT_CLOCK_SPEED_HZ, SPI_TFT_SPI_MODE, cs);

    spi_device_interface_config_t devcfg_disp = {
        .clock_speed_hz = SPI_TFT_CLOCK_SPEED_HZ,
        .mode = 0,
        .spics_io_num = cs, // CS pin
        .queue_size = 1,
        .post_cb = spi_ready,
        .flags = SPI_DEVICE_NO_DUMMY | SPI_DEVICE_HALFDUPLEX,
    };

    // Attach the LCD to the SPI bus
    esp_err_t ret = spi_bus_add_device(host, &devcfg_disp, &spi_disp);
    assert(ret == ESP_OK);
}

void tp_spi_add_device(spi_host_device_t host)
{
    gpio_num_t t_cs;
    esp_err_t ret = gpio_get_touch(&t_cs, NULL);
    if (t_cs == GPIO_NONE)
        return;

    gpio_config_t cs_config = {
        .pin_bit_mask = BIT64(t_cs),
        .mode = GPIO_MODE_OUTPUT,
        .pull_up_en = GPIO_PULLUP_DISABLE,
        .pull_down_en = GPIO_PULLDOWN_DISABLE,
        .intr_type = GPIO_INTR_DISABLE,
    };

    ret = gpio_config(&cs_config);
    assert(ret == ESP_OK);
    spi_device_interface_config_t devcfg_touch = {
        .command_bits = 8,
        .clock_speed_hz = 2000000,
        .mode = 0,
        .spics_io_num = t_cs,
        .queue_size = 1,
        .pre_cb = NULL,
        .post_cb = NULL,
    };

    // Attach the Touch to the SPI bus
    ret = spi_bus_add_device(host, &devcfg_touch, &spi_touch);
    assert(ret == ESP_OK);
}

void disp_spi_send_data(uint8_t *data, uint16_t length)
{
    if (length == 0)
        return;

    while (spi_trans_in_progress)
        ;

    memset(&transaction, 0, sizeof(transaction));
    transaction.length = length * 8; // transaction length is in bits
    transaction.tx_buffer = data;

    spi_trans_in_progress = true;
    spi_color_sent = false;
    ESP_ERROR_CHECK(spi_device_queue_trans(spi_disp, &transaction, portMAX_DELAY));
}

void disp_spi_send_colors(uint8_t *data, uint16_t length)
{
    if (length == 0)
        return;

    while (spi_trans_in_progress)
        ;

    memset(&transaction, 0, sizeof(transaction));
    transaction.length = length * 8; // transaction length is in bits
    transaction.tx_buffer = data;

    spi_trans_in_progress = true;
    spi_color_sent = true;
    ESP_ERROR_CHECK(spi_device_queue_trans(spi_disp, &transaction, portMAX_DELAY));
}

uint16_t tp_spi_xchg_xpt(uint8_t writeData)
{
    while (spi_trans_in_progress)
        ;

    uint8_t data_recv[2] = {0, 0};
    uint16_t data_send = 0;

    memset(&transaction, 0, sizeof(transaction));
    transaction.cmd = writeData;
    transaction.length = 16; // length is in bits
    transaction.tx_buffer = &data_send;
    transaction.rx_buffer = &data_recv;

    ESP_ERROR_CHECK(spi_device_queue_trans(spi_touch, &transaction, portMAX_DELAY));

    spi_transaction_t *rt;
    ESP_ERROR_CHECK(spi_device_get_trans_result(spi_touch, &rt, portMAX_DELAY));

    return (uint16_t)(data_recv[0]) << 8 | (uint16_t)(data_recv[1]);
}

void disp_tp_spi_finished()
{
    while (spi_trans_in_progress)
        ;
}

/**********************
 *   STATIC FUNCTIONS
 **********************/

static void IRAM_ATTR spi_ready(spi_transaction_t *trans)
{

    if (spi_color_sent)
    {
        spi_color_sent = false;
        spi_trans_in_progress = false;
        lv_disp_t *disp = NULL;

        disp = _lv_refr_get_disp_refreshing();
        lv_disp_flush_ready(disp->driver);
    }
    else
    {
        spi_trans_in_progress = false;
    }
}

It is unfortunate that the problem of artifacts on the display screen is of no interest to anyone.

I've encoutered the same issue (with artifacts). I think the artifacts are there, because the last pixel of the buffer didnt get one of the bytes of color sent.
I think the size of mybuf should be 3*size + 1, at least in my case it made the artifacts dissapear.

 uint32_t size = lv_area_get_width(area) * lv_area_get_height(area);

    lv_color16_t *buffer_16bit = (lv_color16_t *) color_map;
    uint8_t *mybuf;
    do {
        mybuf = (uint8_t *) heap_caps_malloc(3 * size * sizeof(uint8_t), MALLOC_CAP_DMA);
        if (mybuf == NULL) {
            LV_LOG_WARN("Could not allocate enough DMA memory!");
        }
    } while (mybuf == NULL);

    uint32_t LD = 0;
    uint32_t j = 0;

    for (uint32_t i = 0; i < size; i++) {
        LD = buffer_16bit[i].full;
        mybuf[j] = (uint8_t) (((LD & 0xF800) >> 8) | ((LD & 0x8000) >> 13));
        j++;
        mybuf[j] = (uint8_t) ((LD & 0x07E0) >> 3);
        j++;
        mybuf[j] = (uint8_t) (((LD & 0x001F) << 3) | ((LD & 0x0010) >> 2));
        j++;
    }